From aslakg at gmail.com  Thu Nov  5 06:58:10 2009
From: aslakg at gmail.com (Aslak Gronflaten)
Date: Thu, 5 Nov 2009 14:58:10 +0800
Subject: [elephant-devel] Upgrading problems
Message-ID: <57870af40911042258x7528abddoe1e144df1a3deaed@mail.gmail.com>

Hi,

I'm (still) having problems upgrading an existing berkeley 4.6 database from
elephant 0.9 to 1.0 trunk, running sbcl under 32 bit linux.

The problem is that I can't access any persistent classes any more.
I believe the data is still there , and I can access btrees for example, but
a class defined like this f.ex:

(DEFPCLASS USER NIL
  ((USERNAME :ACCESSOR USERNAME :INITARG :USERNAME :INITFORM NIL :INDEX T)
   (PASSWORD :ACCESSOR PASSWORD :INITARG :PASSWORD :INITFORM NIL)
   (FULLNAME :ACCESSOR FULLNAME :INITARG :FULLNAME :INITFORM NIL)
   (EMPLOYEE :ACCESSOR EMPLOYEE :INITARG :EMPLOYEE :INITFORM NIL :INDEX T)
   (E-MAIL :ACCESSOR E-MAIL :INITARG :E-MAIL :INITFORM NIL ))
  (:INDEX T))

Pointing to the same db, (get-instances-by-class 'user) will give me a long
list of users under 0.9, while I get an empty list under 1.0.

If I do a describe on the class under 0.9, I get the following:

#<ELEPHANT:PERSISTENT-METACLASS USER>
  [standard-object]

Class precedence-list: USER, ELEPHANT:PERSISTENT-OBJECT,
                       ELEPHANT:PERSISTENT, STANDARD-OBJECT,
                       SB-PCL::SLOT-OBJECT, T
Direct superclasses: ELEPHANT:PERSISTENT-OBJECT
No subclasses.
Direct slots:
  USERNAME
    Allocation: :DATABASE
    Initargs: :USERNAME
    Readers: USERNAME
    Writers: (SETF USERNAME)
  PASSWORD
    Allocation: :DATABASE
    Initargs: :PASSWORD
    Readers: PASSWORD
    Writers: (SETF PASSWORD)
  FULLNAME
    Allocation: :DATABASE
    Initargs: :FULLNAME
    Readers: FULLNAME
    Writers: (SETF FULLNAME)
  EMPLOYEE
    Allocation: :DATABASE
    Initargs: :EMPLOYEE
    Readers: EMPLOYEE
    Writers: (SETF EMPLOYEE)
  E-MAIL
    Allocation: :DATABASE
    Initargs: :E-MAIL
    Readers: E-MAIL
    Writers: (SETF E-MAIL)

Slots with :INSTANCE allocation:
  %TYPE                        = (CLASS #<ELEPHANT:PERSISTENT-METACLASS
USER>)
  SOURCE                       = #S(SB-C:DEFINITION-SOURCE-LOCATION..
  PLIST                        = NIL
  NAME                         = USER
  CLASS-EQ-SPECIALIZER         = #<SB-PCL::CLASS-EQ-SPECIALIZER {B2F4531}>
  DIRECT-SUPERCLASSES          = (#<ELEPHANT:PERSISTENT-METACLASS
ELEPHANT:PERSISTENT-OBJECT>)
  DIRECT-SUBCLASSES            = NIL
  DIRECT-METHODS               = ((#<STANDARD-METHOD IS-IN-PM (USER)
{14CB5FD9}>..
  %DOCUMENTATION               = NIL
  SAFE-P                       = NIL
  FINALIZED-P                  = T
  %CLASS-PRECEDENCE-LIST       = (#<ELEPHANT:PERSISTENT-METACLASS USER>..
  CPL-AVAILABLE-P              = T
  CAN-PRECEDE-LIST             = (#<BUILT-IN-CLASS T> #<SB-PCL::SLOT-CLASS
SB-PCL::SLOT-OBJECT>..
  INCOMPATIBLE-SUPERCLASS-LIST  = NIL
  WRAPPER                      = #<SB-PCL::WRAPPER
#<ELEPHANT:PERSISTENT-METACLASS USER> {AB187A1}>
  PROTOTYPE                    = NIL
  DIRECT-SLOTS                 =
(#<ELEPHANT::PERSISTENT-DIRECT-SLOT-DEFINITION USERNAME>..
  SLOTS                        =
(#<ELEPHANT::TRANSIENT-EFFECTIVE-SLOT-DEFINITION ELEPHANT::%OID>..
  %PERSISTENT-SLOTS            = ((USERNAME PASSWORD FULLNAME EMPLOYEE
E-MAIL))
  %INDEXED-CLASS               = NIL
  %INDEXING-STATE              = (#INDEXING-RECORD<c: (T) islt: 2 dslt: 0>)
  %INDEX-CACHE                 = #<unbound slot>


While doing the same under 1.0 I get this description (top part is
identical):

Slots with :INSTANCE allocation:
  %TYPE                        = (CLASS #<ELEPHANT:PERSISTENT-METACLASS
USER>)
  SOURCE                       = #S(SB-C:DEFINITION-SOURCE-LOCATION..
  PLIST                        = NIL
  NAME                         = USER
  CLASS-EQ-SPECIALIZER         = #<SB-PCL::CLASS-EQ-SPECIALIZER {B163731}>
  DIRECT-SUPERCLASSES          = (#<ELEPHANT:PERSISTENT-METACLASS
ELEPHANT:PERSISTENT-OBJECT>)
  DIRECT-SUBCLASSES            = NIL
  DIRECT-METHODS               = ((#<STANDARD-METHOD IS-IN-PM (USER)
{B7CA389}>..
  %DOCUMENTATION               = NIL
  SAFE-P                       = NIL
  FINALIZED-P                  = T
  %CLASS-PRECEDENCE-LIST       = (#<ELEPHANT:PERSISTENT-METACLASS USER>..
  CPL-AVAILABLE-P              = T
  CAN-PRECEDE-LIST             = (#<BUILT-IN-CLASS T> #<SB-PCL::SLOT-CLASS
SB-PCL::SLOT-OBJECT>..
  INCOMPATIBLE-SUPERCLASS-LIST  = NIL
  WRAPPER                      = #<SB-PCL::WRAPPER
#<ELEPHANT:PERSISTENT-METACLASS USER> {AEB48C1}>
  PROTOTYPE                    = NIL
  DIRECT-SLOTS                 = (#<ELEPHANT::INDEXED-DIRECT-SLOT-DEFINITION
USERNAME>..
  SLOTS                        =
(#<ELEPHANT::TRANSIENT-EFFECTIVE-SLOT-DEFINITION ELEPHANT::OID>..
  %CLASS-SCHEMA                = #<CLASS-SCHEMA USER>
  %STORE-SCHEMAS               = (((:BDB "/www/intranet/db") . #<DB-SCHEMA
34913 USER (s: NIL p: NIL)>))
  %CLASS-INDEXING              = T
  %CACHE-STYLE                 = :NONE


I don't know the internals , so it's just my gut feeling that doesn't like
the s: nil p: nil  there.  Any ideas on how to find what goes wrong?
I desperately need to upgrade as I keep getting Berkeley DB error #-30975:
DB_RUNRECOVERY: Fatal error, run database recovery  in the running system,
after about a days use.
It started out being a monthly occurance, then weekly, and now the system is
really unstable. I'm hoping that getting to 1.0 will help. Also it would
allow upgrading to db4.7.

Any pointers are most welcome....

-- 
Regards Aslak
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/elephant-devel/attachments/20091105/b0287abe/attachment.html>

From sky at viridian-project.de  Thu Nov  5 08:43:37 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Thu, 5 Nov 2009 09:43:37 +0100
Subject: [elephant-devel] Upgrading problems
In-Reply-To: <57870af40911042258x7528abddoe1e144df1a3deaed@mail.gmail.com>
References: <57870af40911042258x7528abddoe1e144df1a3deaed@mail.gmail.com>
Message-ID: <20091105084337.GA22296@viridian-project.de>

On Thu, Nov 05, 2009 at 02:58:10PM +0800, Aslak Gronflaten wrote:

> Any pointers are most welcome....

If all else fails you could use or write a manual import/export
facility.

  Leslie



From sebyte at smolny.plus.com  Thu Nov  5 16:28:25 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Thu, 05 Nov 2009 16:28:25 +0000
Subject: [elephant-devel] memutil.lisp breaks compile
Message-ID: <fx8tgcly.fsf@vps203.linuxvps.org>

Hi there,

Darwin 9.8.0 (MacOSX 10.5.8)
SBCL 1.0.32.10
elephant-1.0 from darcs (altered wnpp-projects)
BerkeleyDB-4.5.20 (installed as per src/contrib/henrik's shell script)

 bash-3.2$ ./clbuild compile elephant
 [...]
 WARNING:
    COMPILE-FILE warned while performing
    #<COMPILE-OP (:VERBOSE NIL) {10032EBFA1}> on
    #<CL-SOURCE-FILE "memutil" {1003E481E1}>.
 unhandled ASDF:COMPILE-FAILED in thread #<SB-THREAD:THREAD "initial thread" RUNNING {100361EDE1}>:
   erred while invoking #<COMPILE-OP (:VERBOSE NIL) {10032EBFA1}> on
   #<CL-SOURCE-FILE "memutil" {1003E481E1}>

 0: (SB-DEBUG::MAP-BACKTRACE #<CLOSURE (LAMBDA #) {1005890E29}>)[:EXTERNAL]
 1: (BACKTRACE 128 #<SYNONYM-STREAM :SYMBOL SB-SYS:*STDERR* {1000157B71}>)
 2: (SB-DEBUG::DEBUGGER-DISABLED-HOOK
     #<ASDF:COMPILE-FAILED {100588DB71}>
     #<unavailable argument>)
 3: (INVOKE-DEBUGGER #<ASDF:COMPILE-FAILED {100588DB71}>)
 4: (ERROR ASDF:COMPILE-FAILED)[:EXTERNAL]
 5: ((SB-PCL::FAST-METHOD ASDF:PERFORM (ASDF:COMPILE-OP ASDF:CL-SOURCE-FILE))
     #<unavailable argument>
     #<unavailable argument>
     #<ASDF:COMPILE-OP (:VERBOSE NIL) {10032EBFA1}>
     #<ASDF:CL-SOURCE-FILE "memutil" {1003E481E1}>)
 6: ((LAMBDA
         (SB-PCL::.PV. SB-PCL::.NEXT-METHOD-CALL. SB-PCL::.ARG0.
          SB-PCL::.ARG1.))
     #<unavailable argument>
     #<unavailable argument>
     #<ASDF:COMPILE-OP (:VERBOSE NIL) {10032EBFA1}>
     #<ASDF:CL-SOURCE-FILE "memutil" {1003E481E1}>)
 7: ((SB-PCL::FAST-METHOD ASDF:PERFORM ASDF:AROUND
      (ASDF:COMPILE-OP ASDF:CL-SOURCE-FILE))
     #<unavailable argument>
     #S(SB-PCL::FAST-METHOD-CALL
        :FUNCTION #<CLOSURE # {100304D209}>
        :PV NIL
        :NEXT-METHOD-CALL NIL
        :ARG-INFO (2))
     #<ASDF:COMPILE-OP (:VERBOSE NIL) {10032EBFA1}>
     #<ASDF:CL-SOURCE-FILE "memutil" {1003E481E1}>)
 8: ((LAMBDA ()))
 9: ((FLET SB-THREAD::WITH-RECURSIVE-LOCK-THUNK))
 10: ((FLET #:WITHOUT-INTERRUPTS-BODY-[CALL-WITH-RECURSIVE-LOCK]291))
 11: (SB-THREAD::CALL-WITH-RECURSIVE-LOCK
      #<CLOSURE (FLET SB-THREAD::WITH-RECURSIVE-LOCK-THUNK) {31FF259}>
      #S(SB-THREAD:MUTEX
         :NAME "World Lock"
         :%OWNER #<SB-THREAD:THREAD "initial thread" RUNNING {100361EDE1}>
         :LUTEX #<unknown pointer object, widetag=#x66 {10007F169F}>))
 12: (SB-C::%WITH-COMPILATION-UNIT #<CLOSURE (LAMBDA #) {1003A6B449}>)[:EXTERNAL]
 13: (ASDF:OPERATE ASDF:LOAD-OP "ele-bdb")[:EXTERNAL]
 14: ((FLET CLBUILD::BUILD))
 15: ((LAMBDA ()))
 16: (SB-INT:SIMPLE-EVAL-IN-LEXENV
      (CLBUILD::WITH-APPLICATION (CLBUILD::PROJECTS-STRING &KEY CLBUILD::DUMP
                                  CLBUILD::VERBOSE CLBUILD::FORCE)
        (LET ((CLBUILD::PROJECTS
               (CL-PPCRE:SPLIT "\\s+" CLBUILD::PROJECTS-STRING))
              (CLBUILD::BLACKLISTED-SYSTEMS
               (UNLESS CLBUILD::FORCE CLBUILD::*BLACKLISTED-SYSTEMS*))
              (CLBUILD::SKIPPED-SYSTEMS 'NIL)
              (*PACKAGE* (FIND-PACKAGE :CL-USER)))
          (SETF CLBUILD::PROJECTS (SORT CLBUILD::PROJECTS #'STRING-LESSP))
          (FLET ((CLBUILD::BUILD ()
                   (DOLIST (CLBUILD::PROJECT CLBUILD::PROJECTS)
                     (DOLIST
                         (CLBUILD::SYSTEM
                          (CLBUILD::PROJECT-TO-SYSTEMS CLBUILD::PROJECT))
                       (COND
                        ((FIND CLBUILD::SYSTEM CLBUILD::BLACKLISTED-SYSTEMS
                               :TEST #'EQUAL)
                         (PUSH CLBUILD::SYSTEM CLBUILD::SKIPPED-SYSTEMS))
                        (T (FORMAT T "Loading ~A...~%" CLBUILD::SYSTEM)
                         (ASDF:OPERATE 'ASDF:LOAD-OP CLBUILD::SYSTEM :VERBOSE
                                       CLBUILD::VERBOSE)))))))
            (IF CLBUILD::VERBOSE
                (CLBUILD::BUILD)
                (HANDLER-BIND ((STYLE-WARNING #'MUFFLE-WARNING)
                               (COMPILER-NOTE #'MUFFLE-WARNING))
                  (LET ((*COMPILE-VERBOSE* NIL)
                        (*COMPILE-PRINT* NIL)
                        (*LOAD-VERBOSE* NIL)
                        (*LOAD-PRINT* NIL))
                    (CLBUILD::BUILD)))))
          (WHEN CLBUILD::SKIPPED-SYSTEMS
            (FORMAT T
                    "WARNING: The following black-listed systems were skipped: ~
                  ~{~A~^, ~}~%~
                 Try --force t to include them.~%"
                    CLBUILD::SKIPPED-SYSTEMS))
          (WHEN CLBUILD::DUMP
            (FORMAT T "Dumping monster.core...~%")
            (FORCE-OUTPUT)
            (SAVE-LISP-AND-DIE "monster.core"))))
      #<NULL-LEXENV>)
 17: (SB-FASL::LOAD-AS-SOURCE
      #<SB-SYS:FD-STREAM for "file /Users/admin/cl/clbuild/clbuild.lisp" {1002D49871}>
      NIL
      NIL)
 18: ((FLET SB-FASL::LOAD-STREAM)
      #<SB-SYS:FD-STREAM for "file /Users/admin/cl/clbuild/clbuild.lisp" {1002D49871}>
      NIL)
 19: (LOAD "/Users/admin/cl/clbuild/clbuild.lisp")[:EXTERNAL]
 20: (SB-INT:SIMPLE-EVAL-IN-LEXENV
      (LOAD "/Users/admin/cl/clbuild/clbuild.lisp")
      #<NULL-LEXENV>)
 21: (SB-IMPL::PROCESS-EVAL/LOAD-OPTIONS
      ((:EVAL . "(require :asdf)")
       (:EVAL
        . "(setq asdf:*central-registry* '(#p\"/Users/admin/cl/clbuild/systems/\"))")
       (:EVAL . "nil")
       (:EVAL . "(load \"/Users/admin/cl/clbuild/clbuild.lisp\")")))
 22: (SB-IMPL::TOPLEVEL-INIT)
 23: ((LABELS SB-IMPL::RESTART-LISP))

 unhandled condition in --disable-debugger mode, quitting

Regards,

Sebastian
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From henrik at evahjelte.com  Thu Nov  5 22:03:06 2009
From: henrik at evahjelte.com (Henrik Hjelte)
Date: Thu, 5 Nov 2009 23:03:06 +0100
Subject: [elephant-devel] memutil.lisp breaks compile
In-Reply-To: <fx8tgcly.fsf@vps203.linuxvps.org>
References: <fx8tgcly.fsf@vps203.linuxvps.org>
Message-ID: <50e8e4f60911051403r4c3b235wc515a956bd4a5952@mail.gmail.com>

On Thu, Nov 5, 2009 at 5:28 PM, Sebastian Tennant
<sebyte at smolny.plus.com> wrote:
> Hi there,
>
> Darwin 9.8.0 (MacOSX 10.5.8)
> SBCL 1.0.32.10
> elephant-1.0 from darcs (altered wnpp-projects)
> BerkeleyDB-4.5.20 (installed as per src/contrib/henrik's shell script)
>
> ?bash-3.2$ ./clbuild compile elephant
> ?[...]
> ?WARNING:
> ? ?COMPILE-FILE warned while performing
> ? ?#<COMPILE-OP (:VERBOSE NIL) {10032EBFA1}> on
> ? ?#<CL-SOURCE-FILE "memutil" {1003E481E1}>.
> ?unhandled ASDF:COMPILE-FAILED in thread #<SB-THREAD:THREAD "initial thread" RUNNING {100361EDE1}>:
> ? erred while invoking #<COMPILE-OP (:VERBOSE NIL) {10032EBFA1}> on
> ? #<CL-SOURCE-FILE "memutil" {1003E481E1}>

If there is a problem compiling it is usually because a version of
cffi pretends to be uffi, and the cffi developers refuses to fix this.
http://www.mail-archive.com/cffi-devel at common-lisp.net/msg01656.html

If this is case, you might want to remind them to fix this on their
mailing list.

If it is not the case, I would double-check the BerkeleyDB version.
But first look for an evil file called uffi.asd but somewhere below
the cffi directory.

/Henrik Hjelte



From aslakg at gmail.com  Fri Nov  6 03:40:43 2009
From: aslakg at gmail.com (Aslak Gronflaten)
Date: Fri, 6 Nov 2009 11:40:43 +0800
Subject: [elephant-devel] Upgrading problems
Message-ID: <57870af40911051940p41dd1140m403cc851d587de44@mail.gmail.com>

Hi,

I've looked into things a bit deeper, and I see that at some point things
changed from having the classes defined in memory be the blueprint, to
having a schema stored in the db itself.
So the question becomes, how do I generate a schema based on my class
definitions, while it not keeping the data which is in the db in 0.9
format...

Regards,
   Aslak
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/elephant-devel/attachments/20091106/9aba3926/attachment.html>

From eslick at media.mit.edu  Sat Nov  7 17:36:48 2009
From: eslick at media.mit.edu (Ian Eslick)
Date: Sat, 7 Nov 2009 09:36:48 -0800
Subject: [elephant-devel] Upgrading problems
In-Reply-To: <57870af40911051940p41dd1140m403cc851d587de44@mail.gmail.com>
References: <57870af40911051940p41dd1140m403cc851d587de44@mail.gmail.com>
Message-ID: <51B0E67D-D66D-425D-9DC5-078CEC800C50@media.mit.edu>

Elephant should automatically create and populate a schema when a  
class definition is evaluated while the DB is open.  I'm surprised it  
doesn't do this when a schema is not available.  I'll look into this.

Ian

On Nov 5, 2009, at 7:40 PM, Aslak Gronflaten wrote:

> Hi,
>
> I've looked into things a bit deeper, and I see that at some point  
> things changed from having the classes defined in memory be the  
> blueprint, to having a schema stored in the db itself.
> So the question becomes, how do I generate a schema based on my  
> class definitions, while it not keeping the data which is in the db  
> in 0.9 format...
>
> Regards,
>    Aslak
> _______________________________________________
> elephant-devel site list
> elephant-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/elephant-devel




From sebyte at smolny.plus.com  Sat Nov  7 22:17:21 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Sat, 07 Nov 2009 22:17:21 +0000
Subject: [elephant-devel] memutil.lisp breaks compile
References: <fx8tgcly.fsf@vps203.linuxvps.org>
	<50e8e4f60911051403r4c3b235wc515a956bd4a5952@mail.gmail.com>
Message-ID: <639mdlou.fsf@vps203.linuxvps.org>

Quoth Henrik Hjelte <henrik at evahjelte.com>:
> http://www.mail-archive.com/cffi-devel at common-lisp.net/msg01656.html

This link appears to be broken (404 Not Found).

Was it intended to point to this conversation:

 http://thread.gmane.org/gmane.lisp.cffi.devel/1720

by chance?

> If there is a problem compiling it is usually because a version of cffi
> pretends to be uffi, and the cffi developers refuses to fix this.

Opinion is divided.  To name but a few:

Jeff Cunningham and Hans H?bner think it should be renamed.

Lu?s Olivera and Daniel Herring think it's best left as it is.

Lu?s is concerned about breaking things that depend on (or at least benefit
from) his kludge (no offence intended).  Daniel Herring makes the point that
hijacking names to provide upgarde/compatibility layers is a time-honoured
tradition.

In my humble opinion, hijacking names works for comprehensive drop-in
replacements based on consensus amongst parties.  CFFI is clearly not a
comprehensive drop-in replacement for UFFI and there's clearly no consensus on
the issue.  I'm therefore in favour of changing the name.

Can't packages that 'use' UFFI (but really use CFFI) be easily updated like so:

 sed -i~ 's/\<uffi\>\(::\?\)/cffi\1/g' *.lisp

> If this is the case, you might want to remind them to fix this on their
> mailing list.

Cross-posted to cffi-devel.

Regarding the original subject, elephant now compiles, but not when using
clbuild's recompile command.

 $ ./clbuild clean elephant
 $ ./clbuild recompile elephant
 [...]
 WARNING:
    COMPILE-FILE warned while performing
    #<COMPILE-OP (:VERBOSE NIL) {10036E8E81}> on
    #<CL-SOURCE-FILE "memutil" {100399B3A1}>.
 unhandled ASDF:COMPILE-FAILED in thread #<SB-THREAD:THREAD "initial thread"
  RUNNING {100361EDE1}>:
   erred while invoking #<COMPILE-OP (:VERBOSE NIL) {10036E8E81}> on
   #<CL-SOURCE-FILE "memutil" {100399B3A1}>

However, if I issue the command again the same warning is issued but about a
different file (serializer2.lisp) so it's nothing to do with memutil.lisp per
se.

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From luismbo at gmail.com  Sun Nov  8 12:26:44 2009
From: luismbo at gmail.com (=?ISO-8859-1?Q?Lu=EDs_Oliveira?=)
Date: Sun, 8 Nov 2009 12:26:44 +0000
Subject: [elephant-devel] [cffi-devel] memutil.lisp breaks compile
In-Reply-To: <639mdlou.fsf@vps203.linuxvps.org>
References: <fx8tgcly.fsf@vps203.linuxvps.org>
	<50e8e4f60911051403r4c3b235wc515a956bd4a5952@mail.gmail.com> 
	<639mdlou.fsf@vps203.linuxvps.org>
Message-ID: <391f79580911080426r2266d9e6ya5c1648c69bae160@mail.gmail.com>

On Sat, Nov 7, 2009 at 10:17 PM, Sebastian Tennant
<sebyte at smolny.plus.com> wrote:
> Lu?s is concerned about breaking things that depend on (or at least benefit
> from) his kludge (no offence intended). ?Daniel Herring makes the point that
> hijacking names to provide upgarde/compatibility layers is a time-honoured
> tradition.
>
> In my humble opinion, hijacking names works for comprehensive drop-in
> replacements based on consensus amongst parties. ?CFFI is clearly not a
> comprehensive drop-in replacement for UFFI and there's clearly no consensus on
> the issue. ?I'm therefore in favour of changing the name.

Patches for CFFI and, most importantly, clbuild are very much welcome. :-)

-- 
Lu?s Oliveira
http://r42.eu/~luis/



From eslick at media.mit.edu  Sun Nov  8 16:48:56 2009
From: eslick at media.mit.edu (Ian Eslick)
Date: Sun, 8 Nov 2009 08:48:56 -0800
Subject: [elephant-devel] Upgrading problems
In-Reply-To: <57870af40911042258x7528abddoe1e144df1a3deaed@mail.gmail.com>
References: <57870af40911042258x7528abddoe1e144df1a3deaed@mail.gmail.com>
Message-ID: <AD265C46-B71B-4975-8685-38004C645987@media.mit.edu>

Did you do a migration/upgrade or just run the old DB under the new  
code base?  -Ian

On Nov 4, 2009, at 10:58 PM, Aslak Gronflaten wrote:

> Hi,
>
> I'm (still) having problems upgrading an existing berkeley 4.6  
> database from
> elephant 0.9 to 1.0 trunk, running sbcl under 32 bit linux.
>
> The problem is that I can't access any persistent classes any more.
> I believe the data is still there , and I can access btrees for  
> example, but
> a class defined like this f.ex:
>
> (DEFPCLASS USER NIL
>   ((USERNAME :ACCESSOR USERNAME :INITARG :USERNAME :INITFORM  
> NIL :INDEX T)
>    (PASSWORD :ACCESSOR PASSWORD :INITARG :PASSWORD :INITFORM NIL)
>    (FULLNAME :ACCESSOR FULLNAME :INITARG :FULLNAME :INITFORM NIL)
>    (EMPLOYEE :ACCESSOR EMPLOYEE :INITARG :EMPLOYEE :INITFORM  
> NIL :INDEX T)
>    (E-MAIL :ACCESSOR E-MAIL :INITARG :E-MAIL :INITFORM NIL ))
>   (:INDEX T))
>
> Pointing to the same db, (get-instances-by-class 'user) will give me  
> a long list of users under 0.9, while I get an empty list under 1.0.
>
> If I do a describe on the class under 0.9, I get the following:
>
> #<ELEPHANT:PERSISTENT-METACLASS USER>
>   [standard-object]
>
> Class precedence-list: USER, ELEPHANT:PERSISTENT-OBJECT,
>                        ELEPHANT:PERSISTENT, STANDARD-OBJECT,
>                        SB-PCL::SLOT-OBJECT, T
> Direct superclasses: ELEPHANT:PERSISTENT-OBJECT
> No subclasses.
> Direct slots:
>   USERNAME
>     Allocation: :DATABASE
>     Initargs: :USERNAME
>     Readers: USERNAME
>     Writers: (SETF USERNAME)
>   PASSWORD
>     Allocation: :DATABASE
>     Initargs: :PASSWORD
>     Readers: PASSWORD
>     Writers: (SETF PASSWORD)
>   FULLNAME
>     Allocation: :DATABASE
>     Initargs: :FULLNAME
>     Readers: FULLNAME
>     Writers: (SETF FULLNAME)
>   EMPLOYEE
>     Allocation: :DATABASE
>     Initargs: :EMPLOYEE
>     Readers: EMPLOYEE
>     Writers: (SETF EMPLOYEE)
>   E-MAIL
>     Allocation: :DATABASE
>     Initargs: :E-MAIL
>     Readers: E-MAIL
>     Writers: (SETF E-MAIL)
>
> Slots with :INSTANCE allocation:
>   %TYPE                        = (CLASS #<ELEPHANT:PERSISTENT- 
> METACLASS USER>)
>   SOURCE                       = #S(SB-C:DEFINITION-SOURCE-LOCATION..
>   PLIST                        = NIL
>   NAME                         = USER
>   CLASS-EQ-SPECIALIZER         = #<SB-PCL::CLASS-EQ-SPECIALIZER  
> {B2F4531}>
>   DIRECT-SUPERCLASSES          = (#<ELEPHANT:PERSISTENT-METACLASS  
> ELEPHANT:PERSISTENT-OBJECT>)
>   DIRECT-SUBCLASSES            = NIL
>   DIRECT-METHODS               = ((#<STANDARD-METHOD IS-IN-PM (USER)  
> {14CB5FD9}>..
>   %DOCUMENTATION               = NIL
>   SAFE-P                       = NIL
>   FINALIZED-P                  = T
>   %CLASS-PRECEDENCE-LIST       = (#<ELEPHANT:PERSISTENT-METACLASS  
> USER>..
>   CPL-AVAILABLE-P              = T
>   CAN-PRECEDE-LIST             = (#<BUILT-IN-CLASS T> #<SB-PCL::SLOT- 
> CLASS SB-PCL::SLOT-OBJECT>..
>   INCOMPATIBLE-SUPERCLASS-LIST  = NIL
>   WRAPPER                      = #<SB-PCL::WRAPPER  
> #<ELEPHANT:PERSISTENT-METACLASS USER> {AB187A1}>
>   PROTOTYPE                    = NIL
>   DIRECT-SLOTS                 = (#<ELEPHANT::PERSISTENT-DIRECT-SLOT- 
> DEFINITION USERNAME>..
>   SLOTS                        = (#<ELEPHANT::TRANSIENT-EFFECTIVE- 
> SLOT-DEFINITION ELEPHANT::%OID>..
>   %PERSISTENT-SLOTS            = ((USERNAME PASSWORD FULLNAME  
> EMPLOYEE E-MAIL))
>   %INDEXED-CLASS               = NIL
>   %INDEXING-STATE              = (#INDEXING-RECORD<c: (T) islt: 2  
> dslt: 0>)
>   %INDEX-CACHE                 = #<unbound slot>
>
>
> While doing the same under 1.0 I get this description (top part is  
> identical):
>
> Slots with :INSTANCE allocation:
>   %TYPE                        = (CLASS #<ELEPHANT:PERSISTENT- 
> METACLASS USER>)
>   SOURCE                       = #S(SB-C:DEFINITION-SOURCE-LOCATION..
>   PLIST                        = NIL
>   NAME                         = USER
>   CLASS-EQ-SPECIALIZER         = #<SB-PCL::CLASS-EQ-SPECIALIZER  
> {B163731}>
>   DIRECT-SUPERCLASSES          = (#<ELEPHANT:PERSISTENT-METACLASS  
> ELEPHANT:PERSISTENT-OBJECT>)
>   DIRECT-SUBCLASSES            = NIL
>   DIRECT-METHODS               = ((#<STANDARD-METHOD IS-IN-PM (USER)  
> {B7CA389}>..
>   %DOCUMENTATION               = NIL
>   SAFE-P                       = NIL
>   FINALIZED-P                  = T
>   %CLASS-PRECEDENCE-LIST       = (#<ELEPHANT:PERSISTENT-METACLASS  
> USER>..
>   CPL-AVAILABLE-P              = T
>   CAN-PRECEDE-LIST             = (#<BUILT-IN-CLASS T> #<SB-PCL::SLOT- 
> CLASS SB-PCL::SLOT-OBJECT>..
>   INCOMPATIBLE-SUPERCLASS-LIST  = NIL
>   WRAPPER                      = #<SB-PCL::WRAPPER  
> #<ELEPHANT:PERSISTENT-METACLASS USER> {AEB48C1}>
>   PROTOTYPE                    = NIL
>   DIRECT-SLOTS                 = (#<ELEPHANT::INDEXED-DIRECT-SLOT- 
> DEFINITION USERNAME>..
>   SLOTS                        = (#<ELEPHANT::TRANSIENT-EFFECTIVE- 
> SLOT-DEFINITION ELEPHANT::OID>..
>   %CLASS-SCHEMA                = #<CLASS-SCHEMA USER>
>   %STORE-SCHEMAS               = (((:BDB "/www/intranet/db") . #<DB- 
> SCHEMA 34913 USER (s: NIL p: NIL)>))
>   %CLASS-INDEXING              = T
>   %CACHE-STYLE                 = :NONE
>
>
> I don't know the internals , so it's just my gut feeling that  
> doesn't like the s: nil p: nil  there.  Any ideas on how to find  
> what goes wrong?
> I desperately need to upgrade as I keep getting Berkeley DB error  
> #-30975: DB_RUNRECOVERY: Fatal error, run database recovery  in the  
> running system, after about a days use.
> It started out being a monthly occurance, then weekly, and now the  
> system is really unstable. I'm hoping that getting to 1.0 will help.  
> Also it would allow upgrading to db4.7.
>
> Any pointers are most welcome....
>
> -- 
> Regards Aslak
> _______________________________________________
> elephant-devel site list
> elephant-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/elephant-devel




From sebyte at smolny.plus.com  Mon Nov  9 19:49:38 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Mon, 09 Nov 2009 19:49:38 +0000
Subject: [elephant-devel] Docs in Info format
Message-ID: <ljiftr59.fsf@vps203.linuxvps.org>

Hi there,

I am able to build a near fully-functional Info manual by running 'make
includes-stuff', visiting elephant.texinfo in emacs, updating all nodes,
rebuilding the master menu and removing the html conditional that surrounds it.
I can then run 'makeinfo --force elephant.texinfo' with only 3 warnings [1] and
pretty decent results.

How about an info target in the Makefile, or a README describing how it's done
at least?

Regards,

Sebastian

[1] /user-guide.texinfo:1073: warning: `.' or `,' must follow @xref, not `f'.
    /copying.texinfo:3: Next field of node `Copyright and License' not pointed to
     (perhaps incorrect sectioning?).
    /elephant.texinfo:124: This node (Concept Index) has the bad Prev.

-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sebyte at smolny.plus.com  Mon Nov  9 21:51:41 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Mon, 09 Nov 2009 21:51:41 +0000
Subject: [elephant-devel] "file is not of required architecture"
Message-ID: <fx8ntlhu.fsf@vps203.linuxvps.org>

Hi list,

elephant-1.0            (git)
SBCL                    (git)
Intel Core 2 Duo        MacOS X Leopard

I built and installed BerkeleyDB 4.5 as per /src/contrib/henrik's script.  It
compiled prefectly cleanly, withtout any problems at all.

my-config.sexp reads:

 ((:compiler . :gcc)
  (:berkeley-db-version . "4.5")
  (:berkeley-db-include-dir . "/usr/local/BerkeleyDB.4.5/include/")
  (:berkeley-db-lib-dir . "/usr/local/BerkeleyDB.4.5/lib/")
  (:berkeley-db-lib . "/usr/local/BerkeleyDB.4.5/lib/libdb-4.5.dylib")
  (:berkeley-db-deadlock . "/usr/local/BerkeleyDB.4.5/bin/db_deadlock")
  (:berkeley-db-cachesize . 20971520)
  (:berkeley-db-max-locks . 2000)
  (:berkeley-db-max-objects . 2000)
  (:berkeley-db-map-degree2 . t)
  (:berkeley-db-mvcc . nil)
  (:clsql-lib-paths . nil)
  (:prebuilt-libraries . nil))

I get the following error when I try and open a store:

 (asdf:operate 'asdf:load-op 'elephant)
 (use-package :elephant)
 (defconstant +store+ '(:BDB #p"/path/to/my/store/"))
 (open-store +store+)

 ;;; => erred while invoking #<COMPILE-OP NIL {1003FF3A01}> on
 ;;;    #<BDB-C-SOURCE "libberkeley-db" {1003FDF6F1}>
 ;;;       [Condition of type ASDF:OPERATION-ERROR]

Full text of error:

 Attempting to load libmemutil.dylib...
 Loaded /Users/admin/cl/clbuild/source/elephant/src/memutil/libmemutil.dylib
 ; $ gcc -L/usr/local/BerkeleyDB.4.5/lib/ -I/usr/local/BerkeleyDB.4.5/include/ -bundle -arch x86_64 -fPIC -Wall -g -O2 -g /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c -o
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.dylib -lm -ldb-4.5
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c: In function ?lisp_compare2?:
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c:1061: warning: unused variable ?i?
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c: In function ?lisp_compare_key2?:
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c:1155: warning: unused variable ?i?
 ld warning: in /usr/local/BerkeleyDB.4.5/lib//libdb-4.5.dylib, file is not of required architecture
 Undefined symbols:                                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
   "_db_env_create", referenced from:
       _db_env_cr in cc1XG9uq.o
   "_db_strerror", referenced from:
       _db_strerr in cc1XG9uq.o
   "_db_create", referenced from:
       _db_cr in cc1XG9uq.o
   "_db_sequence_create", referenced from:
       _db_sequence_create2 in cc1XG9uq.o
 ld: symbol(s) not found
 collect2: ld returned 1 exit status
 erred while invoking #<COMPILE-OP NIL {1003FF3A01}> on
 #<BDB-C-SOURCE "libberkeley-db" {1003FDF6F1}>
    [Condition of type ASDF:OPERATION-ERROR]

Inspection shows:

 $ file /usr/local/BerkeleyDB.4.5/lib/libdb-4.5.dylib
 libdb-4.5.dylib: Mach-O dynamically linked shared library i386

But:

 $ file ~/clbuild/source/elephant/src/memutil/libmemutil.dylib
 libmemutil.dylib: Mach-O 64-bit bundle x86_64

so I wonder if there's a way to tell BDB to build 64-bit dynamic libraries, or
if that's even the problem.

Any help/tips/pointers much appreciated.

For instance, how to test my BerkeleyDB installation without using elephant?

Regards,

Seb
--
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sky at viridian-project.de  Tue Nov 10 08:36:08 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Tue, 10 Nov 2009 09:36:08 +0100 (CET)
Subject: [elephant-devel] "file is not of required architecture"
In-Reply-To: <fx8ntlhu.fsf@vps203.linuxvps.org>
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
Message-ID: <96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>


Sebastian Tennant wrote:
> Hi list,
>
> elephant-1.0            (git)
> SBCL                    (git)
> Intel Core 2 Duo        MacOS X Leopard
>
> I built and installed BerkeleyDB 4.5 as per /src/contrib/henrik's script.  It
> compiled prefectly cleanly, withtout any problems at all.

Is there a specific reason you're not using 4.7?


>  $ file /usr/local/BerkeleyDB.4.5/lib/libdb-4.5.dylib
>  libdb-4.5.dylib: Mach-O dynamically linked shared library i386
>
> But:
>
>  $ file ~/clbuild/source/elephant/src/memutil/libmemutil.dylib
>  libmemutil.dylib: Mach-O 64-bit bundle x86_64
>
> so I wonder if there's a way to tell BDB to build 64-bit dynamic libraries, or
> if that's even the problem.

Yes, I think that's your problem. Try

  export CFLAGS=-march=x86_64

before running configure and make for BDB.
Or just try your operating system's BDB package.


> For instance, how to test my BerkeleyDB installation without using elephant?

There's a test suite in the BDB source tree. Or you could write a small test
program in C.

  Leslie

-- 
http://www.linkedin.com/in/polzer




From sky at viridian-project.de  Tue Nov 10 08:38:12 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Tue, 10 Nov 2009 09:38:12 +0100 (CET)
Subject: [elephant-devel] Docs in Info format
In-Reply-To: <ljiftr59.fsf@vps203.linuxvps.org>
References: <ljiftr59.fsf@vps203.linuxvps.org>
Message-ID: <3598cdf1a1f943b422bcbf4ec8e6568e.squirrel@mail.stardawn.org>


Sebastian Tennant wrote:

> How about an info target in the Makefile, or a README describing how it's done
> at least?

I'm positive about that, got a patch?

  Leslie

-- 
http://www.linkedin.com/in/polzer




From sebyte at smolny.plus.com  Tue Nov 10 09:15:52 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Tue, 10 Nov 2009 09:15:52 +0000
Subject: [elephant-devel] Docs in Info format
References: <ljiftr59.fsf@vps203.linuxvps.org>
	<3598cdf1a1f943b422bcbf4ec8e6568e.squirrel@mail.stardawn.org>
Message-ID: <skcmagfr.fsf@vps203.linuxvps.org>

Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
> Sebastian Tennant wrote:
>
>> How about an info target in the Makefile, or a README describing how it's done
>> at least?
>
> I'm positive about that, got a patch?

I'll look into it.  

First off, is there a good reason why elephant.texinfo is shipped the way it is
(with the master menu in need of rebuilding and wrapped in an @ifhtml
conditional)?

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sebyte at smolny.plus.com  Tue Nov 10 09:21:37 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Tue, 10 Nov 2009 09:21:37 +0000
Subject: [elephant-devel] "file is not of required architecture"
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
Message-ID: <my2uag66.fsf@vps203.linuxvps.org>

Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
> Is there a specific reason you're not using 4.7?

Yes there is.  It clearly states in the manual that elephant only works with
BerkeleyDB version 4.5.

 (info "(elephant)Berkeley DB")
 http://common-lisp.net/project/elephant/doc/Berkeley-DB.html#Berkeley-DB

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sky at viridian-project.de  Tue Nov 10 10:44:33 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Tue, 10 Nov 2009 11:44:33 +0100
Subject: [elephant-devel] "file is not of required architecture"
In-Reply-To: <my2uag66.fsf@vps203.linuxvps.org>
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
Message-ID: <20091110104433.GA19646@viridian-project.de>

On Tue, Nov 10, 2009 at 09:21:37AM +0000, Sebastian Tennant wrote:
> Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
> > Is there a specific reason you're not using 4.7?
> 
> Yes there is.  It clearly states in the manual that elephant only works with
> BerkeleyDB version 4.5.
> 
>  (info "(elephant)Berkeley DB")
>  http://common-lisp.net/project/elephant/doc/Berkeley-DB.html#Berkeley-DB

This passage is terribly outdated. Use BDB 4.7.

  Leslie



From sebyte at smolny.plus.com  Tue Nov 10 18:35:38 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Tue, 10 Nov 2009 18:35:38 +0000
Subject: [elephant-devel] "file is not of required architecture"
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
Message-ID: <hbt28byd.fsf@vps203.linuxvps.org>

Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
> This passage is terribly outdated. Use BDB 4.7.

Well I'll update the passage when I submit my patch.

In the mean time, it'd be nice to get elephant working on my Mac (I bet it's
fine under Debian).

I've installed MacPort's db47 but am getting the same error as before:

 ; $ gcc -L/opt/local/lib/db47/ -I/opt/local/include/db47/ -bundle -arch x86_64 -fPIC -Wall -g -O2 -g
 /Users/admin/cl/clbuild/source\ /elephant/src/db-bdb/libberkeley-db.c -o /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.dylib -lm -ldb-4.7
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c: In function ?lisp_compare2?:
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c:1061: warning: unused variable ?i?
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c: In function ?lisp_compare_key2?:
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c:1155: warning: unused variable ?i?
 ld warning: in /opt/local/lib/db47//libdb-4.7.dylib, file is not of required architecture
 Undefined symbols:                                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
   "_db_env_create", referenced from:
       _db_env_cr in ccmOlpXz.o
   "_db_strerror", referenced from:
       _db_strerr in ccmOlpXz.o
   "_db_create", referenced from:
       _db_cr in ccmOlpXz.o
   "_db_sequence_create", referenced from:
       _db_sequence_create2 in ccmOlpXz.o
 ld: symbol(s) not found
 collect2: ld returned 1 exit status
 erred while invoking #<COMPILE-OP NIL {1004908A01}> on
 #<BDB-C-SOURCE "libberkeley-db" {10048FAC41}>
    [Condition of type ASDF:OPERATION-ERROR]

my-config.sexp now reads:

 ((:compiler . :gcc)
  (:berkeley-db-version . "4.5")
  (:berkeley-db-include-dir . "/opt/local/include/db47/")
  (:berkeley-db-lib-dir . "/opt/local/lib/db47/")
  (:berkeley-db-lib . "/opt/local/lib/db47/libdb-4.7.dylib")
  (:berkeley-db-deadlock . "/opt/local/bin/db47_deadlock")
  (:berkeley-db-cachesize . 20971520)
  (:berkeley-db-max-locks . 2000)
  (:berkeley-db-max-objects . 2000)
  (:berkeley-db-map-degree2 . t)
  (:berkeley-db-mvcc . nil)
  (:clsql-lib-paths . nil)
  (:prebuilt-libraries . nil))

The commands the lead to the fail:

 (defconstant +store+
   '(:BDB #p"/Users/admin/cl/my/test/store/")) ;dir created in advnace
 (open-store +store+) ;this fails

Any help very much appreciated.

Seb
--
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sky at viridian-project.de  Tue Nov 10 19:17:56 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Tue, 10 Nov 2009 20:17:56 +0100
Subject: [elephant-devel] "file is not of required architecture"
In-Reply-To: <hbt28byd.fsf@vps203.linuxvps.org>
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
Message-ID: <20091110191756.GA6201@viridian-project.de>

On Tue, Nov 10, 2009 at 06:35:38PM +0000, Sebastian Tennant wrote:

> I've installed MacPort's db47 but am getting the same error as before:

Are you sure that it's compiled for 64bit?



From sebyte at smolny.plus.com  Thu Nov 12 13:48:17 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Thu, 12 Nov 2009 13:48:17 +0000
Subject: [elephant-devel] "file is not of required architecture"
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
Message-ID: <7htv7t26.fsf@vps203.linuxvps.org>

Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
> On Tue, Nov 10, 2009 at 06:35:38PM +0000, Sebastian Tennant wrote:
>> I've installed MacPort's db47 but am getting the same error as before:
> Are you sure that it's compiled for 64bit?

I've no doubt that's what the problem is.

My Mac is a 64-bit machine but installing BerkeleyDB 4.7.X (db47) using
MacPorts appears not to take advantage of its 64-bit capability:

 $ file /opt/local/lib/db47/libdb-4.7.dylib
 Mach-O dynamically linked shared library i386

But when elephant is compiled, it definitely builds 64-bit dynamic libraries:

 $ file /Users/admin/cl/clbuild/source/elephant/src/memutil/libmemutil.dylib
 Mach-O 64-bit bundle x86_64

Do you know how to either force a 64-bit build of BerkeleyDB 4.7 (from source)
or force elephant to compile a 32-bit libmemutil.dylib?

Regards,

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From rsynnott at gmail.com  Thu Nov 12 13:51:39 2009
From: rsynnott at gmail.com (Robert Synnott)
Date: Thu, 12 Nov 2009 13:51:39 +0000
Subject: [elephant-devel] "file is not of required architecture"
In-Reply-To: <7htv7t26.fsf@vps203.linuxvps.org>
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
Message-ID: <24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>

If you install BDB from Oracle's source, you might get a 64bit build
(or have the option to ask for one in the configuration stage). I
gather  that MacPorts hasn't been adapted too well for Snow Leopard as
yet.
Rob

2009/11/12 Sebastian Tennant <sebyte at smolny.plus.com>:
> Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
>> On Tue, Nov 10, 2009 at 06:35:38PM +0000, Sebastian Tennant wrote:
>>> I've installed MacPort's db47 but am getting the same error as before:
>> Are you sure that it's compiled for 64bit?
>
> I've no doubt that's what the problem is.
>
> My Mac is a 64-bit machine but installing BerkeleyDB 4.7.X (db47) using
> MacPorts appears not to take advantage of its 64-bit capability:
>
> ?$ file /opt/local/lib/db47/libdb-4.7.dylib
> ?Mach-O dynamically linked shared library i386
>
> But when elephant is compiled, it definitely builds 64-bit dynamic libraries:
>
> ?$ file /Users/admin/cl/clbuild/source/elephant/src/memutil/libmemutil.dylib
> ?Mach-O 64-bit bundle x86_64
>
> Do you know how to either force a 64-bit build of BerkeleyDB 4.7 (from source)
> or force elephant to compile a 32-bit libmemutil.dylib?
>
> Regards,
>
> Seb
> --
> Emacs' AlsaPlayer - Music Without Jolts
> Lightweight, full-featured and mindful of your idyllic happiness.
> http://home.gna.org/eap
>
>
> _______________________________________________
> elephant-devel site list
> elephant-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/elephant-devel
>



-- 
Robert Synnott
http://myblog.rsynnott.com
MSN: rsynnott at gmail.com
Jabber: rsynnott at gmail.com



From sebyte at smolny.plus.com  Thu Nov 12 15:23:12 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Thu, 12 Nov 2009 15:23:12 +0000
Subject: [elephant-devel] "file is not of required architecture"
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
	<24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>
Message-ID: <3a4j7onz.fsf@vps203.linuxvps.org>

Quoth Robert Synnott <rsynnott at gmail.com>:
> If you install BDB from Oracle's source, you might get a 64bit build (or have
> the option to ask for one in the configuration stage). I gather that MacPorts
> hasn't been adapted too well for Snow Leopard as yet.

Noted, although I'm not actually running Snow Leopard.

Any idea what the magic incantation to ./compile is, or which file I should
look in to find out?

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sebyte at smolny.plus.com  Thu Nov 12 15:57:50 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Thu, 12 Nov 2009 15:57:50 +0000
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian box
	either
Message-ID: <fx8jrb0h.fsf@vps203.linuxvps.org>

Hi list,

I'm disappointed to have to say that elephant's BDB store doesn't work
out-of-the-box on my 32-bit x86 Debian box either:

 Attempting to load libmemutil.so...
 Loaded /home/sebyte/lisp/cl/clbuild/source/elephant/src/memutil/libmemutil.so
 Loading /usr/local/BerkeleyDB.4.7/lib/libdb-4.7.so
 Attempting to load libberkeley-db.so...
 Loaded
 /home/sebyte/lisp/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.so
 ;
 ; compilation unit aborted
 ;   caught 1 fatal ERROR condition

 Store controller init function not registered for data store BDB.
  [Condition of type SIMPLE-ERROR]

A user complained about the same error on March 14th this year on #lisp but the
solution, if any, is not readily available:

 http://rotty.yi.org/irclogs/freenode/%23lisp/2009-03-14/#e205

I'm using elephant-1.0 from darcs, BerkeleyDB 4.7 from Oracle (built from
source) and SBCL 1.0.32.14.

The code that cause the error:

 (defconstant +store+ '(:BDB "/home/sebyte/lisp/cl/my/store/"))
 (elephant:open-store +store+) ; => ERROR

Any help/tips/pointers very much appreciated.

Regards,

Seb
--
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From rme at clozure.com  Thu Nov 12 16:47:55 2009
From: rme at clozure.com (R. Matthew Emerson)
Date: Thu, 12 Nov 2009 11:47:55 -0500
Subject: [elephant-devel] "file is not of required architecture"
In-Reply-To: <3a4j7onz.fsf@vps203.linuxvps.org>
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
	<24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>
	<3a4j7onz.fsf@vps203.linuxvps.org>
Message-ID: <718A5688-F55B-49D6-8371-243B1358AF46@clozure.com>


On Nov 12, 2009, at 10:23 AM, Sebastian Tennant wrote:

> Quoth Robert Synnott <rsynnott at gmail.com>:
>> If you install BDB from Oracle's source, you might get a 64bit build (or have
>> the option to ask for one in the configuration stage). I gather that MacPorts
>> hasn't been adapted too well for Snow Leopard as yet.
> 
> Noted, although I'm not actually running Snow Leopard.
> 
> Any idea what the magic incantation to ./compile is, or which file I should
> look in to find out?

1. get bdb 4.7

Direct link; if it doesn't work use the index page:
http://download.oracle.com/berkeley-db/db-4.7.25.tar.gz

Index page:
http://www.oracle.com/technology/software/products/berkeley-db/db/index.html

2. unpack into /usr/local/src (or wherever)

Build docs are at

file:///usr/local/src/db-4.7.25/docs/ref/build_unix/intro.html

(or wherever you put it)

3. cd to db-4.7.25/build_unix

4. use the following command line to configure, where $ is the shell prompt

$ CFLAGS="-arch x86_64" LDFLAGS="-arch x86_64" ../dist/configure


Then proceed normally (e.g., make, then sudo make install).



From sebyte at smolny.plus.com  Thu Nov 12 17:31:35 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Thu, 12 Nov 2009 17:31:35 +0000
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian box
	either
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
Message-ID: <3a4jr6o8.fsf@vps203.linuxvps.org>

Quoth Sebastian Tennant <sebyte at smolny.plus.com>:
>  Store controller init function not registered for data store BDB.
>   [Condition of type SIMPLE-ERROR]

Scratch the above error - I can no longer reproduce it myself since deciding to
delete all relevant fasl files, and starting from scratch.

 CL-USER> (defconstant +store+ '(:BDB "/home/sebyte/lisp/cl/my/store/"))
 +STORE+
 CL-USER> (elephant:open-store +store+)
 [...]
 Attempting to load libmemutil.so...
 Loaded /home/sebyte/lisp/cl/clbuild/source/elephant/src/memutil/libmemutil.so

 ; compiling file
   "/home/sebyte/lisp/cl/clbuild/source/elephant/src/db-bdb/package.lisp"
   (written 11 NOV 2009 09:51:21 PM):
 ; compiling (IN-PACKAGE :CL-USER)
 ; compiling (DEFPACKAGE SLEEPYCAT ...)
 ; compiling (DEFPACKAGE DB-BDB ...)

 ; /home/sebyte/lisp/cl/clbuild/source/elephant/src/db-bdb/package.fasl written
 ; compilation finished in 0:00:00.064

At this point sbcl hangs.

If don't bother to remove all the fasls again, restart lisp and reload elephant
I get an actual error that's handled by sldb (rather than a hang):

 CL-USER> (defconstant +store+ '(:BDB "/home/sebyte/lisp/cl/my/store/"))
 +STORE+
 CL-USER> (elephant:open-store +store+)
 [...]
 Attempting to load libmemutil.so...
 Loaded /home/sebyte/lisp/cl/clbuild/source/elephant/src/memutil/libmemutil.so
 Loading /usr/local/BerkeleyDB.4.7/lib/libdb-4.7.so
 Attempting to load libberkeley-db.so...
 Loaded
 ;     /home/sebyte/lisp/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.so

 end of file on #<SB-SYS:FD-STREAM for "file
  /home/sebyte/lisp/cl/clbuild/source/elephant/src/db-bdb/berkeley-constants" {BB13669}>
    [Condition of type END-OF-FILE]                       ^^^^^^^^^^^^^^^^^^

If I repeat it a thrid time I get the same error, so there's definitely a
problem with berkeley-constants.

For the last time:

 Debian Lenny, i386 32-bit
 SBCL 1.0.32.14
 elephant-1.0 from darcs
 BDB 4.7
 All other dependencies up-to-date as far as clbuild is concerned.

Would an elephant developer please help.  (I'm assuming you're not a developer
Leslie). I really want to start actually using elephant!!!.

I'm starting to wonder if many people really use elephant, and whether any
elephant developers actually read this list?

In short, I'm starting to become pretty disillusioned with elephant in
particular and the state of Common Lisp libraries generally.

Is it because I'm using bleeding edge versions of everything I wonder?  

Seb
--
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sebyte at smolny.plus.com  Thu Nov 12 17:38:55 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Thu, 12 Nov 2009 17:38:55 +0000
Subject: [elephant-devel] "file is not of required architecture"
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
	<24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>
	<3a4j7onz.fsf@vps203.linuxvps.org>
	<718A5688-F55B-49D6-8371-243B1358AF46@clozure.com>
Message-ID: <y6mbprrk.fsf@vps203.linuxvps.org>

Quoth "R. Matthew Emerson" <rme at clozure.com>:
>> Any idea what the magic incantation to ./compile is, or which file I should
>> look in to find out?

> $ CFLAGS="-arch x86_64" LDFLAGS="-arch x86_64" ../dist/configure

Ah!!! Thank you.  They're the magic flags I was looking for.

Now let's see if I can get elephant's bdb store working on my Mac.

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sky at viridian-project.de  Thu Nov 12 18:36:08 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Thu, 12 Nov 2009 19:36:08 +0100
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian
	box	either
In-Reply-To: <3a4jr6o8.fsf@vps203.linuxvps.org>
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
Message-ID: <20091112183608.GA309@viridian-project.de>

On Thu, Nov 12, 2009 at 05:31:35PM +0000, Sebastian Tennant wrote:
> For the last time:
> 
>  Debian Lenny, i386 32-bit
>  SBCL 1.0.32.14
>  elephant-1.0 from darcs
>  BDB 4.7
>  All other dependencies up-to-date as far as clbuild is concerned.
> 
> Would an elephant developer please help.  (I'm assuming you're not a developer
> Leslie). I really want to start actually using elephant!!!.

I guess you could say there are two somewhat active Elephant developers
right now, and I'm one of them.

 
> I'm starting to wonder if many people really use elephant,

Ian and me are using it every day in GNU/Linux environments..


> and whether any elephant developers actually read this list?
> 
> In short, I'm starting to become pretty disillusioned with elephant in
> particular and the state of Common Lisp libraries generally.

I'm sorry to hear this, and I can't really agree. It's true that
sometimes stuff doesn't work out of the box, but that's the case
everywhere unless the project is *very* well-staffed and used
by a large number of people at the same time.

 
> Is it because I'm using bleeding edge versions of everything I wonder?  

Please, be patient. This is unpaid support and from what I've gathered
so far your questions on this and other lists have all been answered
within a few days maximum.

People's systems can differ vastly (e.g. libraries, bus width, Lisp
implementation) and not all of these issues can be figured out
immediately, no matter how competent the person at the other end is.

  Leslie



From sebyte at smolny.plus.com  Fri Nov 13 08:32:33 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Fri, 13 Nov 2009 08:32:33 +0000
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian
	box	either
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
Message-ID: <ljia7rku.fsf@vps203.linuxvps.org>

Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
>> Would an elephant developer please help.  (I'm assuming you're not a developer
>> Leslie). I really want to start actually using elephant!!!.

> I guess you could say there are two somewhat active Elephant developers right
> now, and I'm one of them.

I'm sorry Leslie.  It's just that you're not listed under 'Authors' at:

 http://common-lisp.net/project/elephant/about.html

Ian (Eslick) is.

>> I'm starting to wonder if many people really use elephant,

> Ian and me are using it every day in GNU/Linux environments..

Two isn't many :)

>> In short, I'm starting to become pretty disillusioned with elephant in
>> particular and the state of Common Lisp libraries generally.

> I'm sorry to hear this, and I can't really agree. It's true that
> sometimes stuff doesn't work out of the box, but that's the case
> everywhere unless the project is *very* well-staffed and used
> by a large number of people at the same time.

You're right.  I suppose what I'm really saying is that it's a shame a large
number of people aren't using these tools.

>> Is it because I'm using bleeding edge versions of everything I wonder?  

> Please, be patient. This is unpaid support and from what I've gathered
> so far your questions on this and other lists have all been answered
> within a few days maximum.

Again you are right.  I have had responses within a day or two at most and I'm
very grateful for every response I receive.  I also fully appreciate that no
one is paid for helping me, or anyone else, with their problems.

My experience with elephant has been particularly trying and my frustration got
the better of me last night, that's all.

> People's systems can differ vastly (e.g. libraries, bus width, Lisp
> implementation) and not all of these issues can be figured out
> immediately, no matter how competent the person at the other end is.

Understood, however my two systems (clbuild/64-bit/SBCL/MacOSX and
clbuild/32-bit/SBCL/Debian Lenny) are pretty common platforms are they not?

What's more, it's not obscure corner/edge cases which are causing me problems -
I haven't been able to get elephant working sufficiently well to even begin
evaluating it for my needs.

Anyway, moaning is pointless.  Action is what's needed.  A few prominent
notices on the elephant's website would have saved me hours, if not days.

 1) If you use clbuild - edit wnpp-projects so that elephant-1.0 is pulled from
    darcs.
 2) Install UFFI (available from URL) and ensure that the 'real' uffi.asd is
    found by ASDF first.
 3) If you want to use a BerkeleyDB store, disregard the manual and install
    Berkeley DB version 4.7 (from URL).  Better still, update the manual.
 4) If you have a 64-bit machine make sure you compile a 64-bit BerkeleyDB.
    Elephant will expect it. Include instructions for said compilation.

There are probably other things too, but I can't think of them at the moment.
I'm happy to do these things myself.  I'll post a patched elephant.texinfo
here, but what about the rest of the above notices?  Where do you think they
should go?  In a README, or on the website perhaps?

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sky at viridian-project.de  Fri Nov 13 09:19:31 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Fri, 13 Nov 2009 10:19:31 +0100 (CET)
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian
 box	either
In-Reply-To: <ljia7rku.fsf@vps203.linuxvps.org>
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
Message-ID: <cd587bcc15f4024e7e9dd518f304abe4.squirrel@mail.stardawn.org>


Sebastian Tennant wrote:
> Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
>>> Would an elephant developer please help.  (I'm assuming you're not a developer
>>> Leslie). I really want to start actually using elephant!!!.
>
>> I guess you could say there are two somewhat active Elephant developers right
>> now, and I'm one of them.
>
> I'm sorry Leslie.  It's just that you're not listed under 'Authors' at:
>
>  http://common-lisp.net/project/elephant/about.html
>
> Ian (Eslick) is.

I don't know much about this. My guess is that this page only lists
authors that have committed a chunk big enough to be relevant in
copyright matters.

I've contributed a substantial number of fixes and help people when
they face problems but I don't have the time to develop new features.


>> Ian and me are using it every day in GNU/Linux environments..
>
> Two isn't many :)

Well there are more but I don't know what use they make of Elephant
so I haven't mentioned them.


> Understood, however my two systems (clbuild/64-bit/SBCL/MacOSX and
> clbuild/32-bit/SBCL/Debian Lenny) are pretty common platforms are they not?

It's well known that Snow Leopard breaks a huge lot of MacPorts stuff,
unfortunately (you can tell from that I'm not even a Mac user to begin
with).

I agree that the second platform you've mentioned is pretty common but
my statement was to be understood in a general case. I.e. we have to
tend to a lot of software combinations -- this doesn't necessarily
mean that it will automatically work on every one of those even
if it's a common one.


> What's more, it's not obscure corner/edge cases which are causing me problems -
> I haven't been able to get elephant working sufficiently well to even begin
> evaluating it for my needs.

In general:

Elephant does funky things like building a shared library in its build process,
stressing the MOP heavily and using UFFI internals that don't cope well with the
UFFI emulation of CFFI.

It doesn't help of course that clbuild hasn't agreed to provide the real UFFI
for Elephant so far.


> Anyway, moaning is pointless.  Action is what's needed.  A few prominent
> notices on the elephant's website would have saved me hours, if not days.

Thanks for being constructive. I agree that the next objective should be
avoiding that new users fall into the same traps over and over.

I already guessed that you were just pretty exhausted and disappointed
after hours of struggling so I didn't really resent your complaints.

Sorry that it didn't work well for you right away.


>  1) If you use clbuild - edit wnpp-projects so that elephant-1.0 is pulled from
>     darcs.
>  2) Install UFFI (available from URL) and ensure that the 'real' uffi.asd is
>     found by ASDF first.
>  3) If you want to use a BerkeleyDB store, disregard the manual and install
>     Berkeley DB version 4.7 (from URL).  Better still, update the manual.
>  4) If you have a 64-bit machine make sure you compile a 64-bit BerkeleyDB.
>     Elephant will expect it. Include instructions for said compilation.

So does it work for you now or do you still face that obscure problem
with bdb-constants.lisp which looks like that file was truncated on your
installation?


> There are probably other things too, but I can't think of them at the moment.
> I'm happy to do these things myself.  I'll post a patched elephant.texinfo
> here, but what about the rest of the above notices?  Where do you think they
> should go?  In a README, or on the website perhaps?

This seems to be the right place as it's referenced from "Downloads":

  http://common-lisp.net/project/elephant/doc/Installation.html#Installation

 Leslie




From sebyte at smolny.plus.com  Fri Nov 13 09:23:29 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Fri, 13 Nov 2009 09:23:29 +0000
Subject: [elephant-devel] "file is not of required architecture"
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
	<24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>
	<3a4j7onz.fsf@vps203.linuxvps.org>
	<718A5688-F55B-49D6-8371-243B1358AF46@clozure.com>
	<y6mbprrk.fsf@vps203.linuxvps.org>
Message-ID: <hbsy7p7y.fsf@vps203.linuxvps.org>

Quoth Sebastian Tennant <sebyte at smolny.plus.com>:
> Now let's see if I can get elephant's bdb store working on my Mac.

Well, I'm a lot closer, but still not quite there yet.

I compiled a 64-bit BerkeleyDB but now I get a 'Berkeley DB error #22: Invalid
argument' error message when I try to (open-store ...):

 bash-3.2$ cat elephant_fail.txt | grep -v '^\;\|^$'
 Attempting to load libmemutil.dylib...
 Loaded /Users/admin/cl/clbuild/source/elephant/src/memutil/libmemutil.dylib
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c: In function ?lisp_compare2?:
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c:1061: warning: unused variable ?i?
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c: In function ?lisp_compare_key2?:
 /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c:1155: warning: unused variable ?i?
 Loading /usr/local/BerkeleyDB.4.7/lib/libdb-4.7.dylib
 Attempting to load libberkeley-db.dylib...
 Loaded /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.dylib
 STYLE-WARNING: Implicitly creating new generic function START-DEADLOCK-DETECTOR.
 STYLE-WARNING: Implicitly creating new generic function STOP-DEADLOCK-DETECTOR.
 STYLE-WARNING: Implicitly creating new generic function CHECKPOINT.
 STYLE-WARNING: Implicitly creating new generic function BUILD-BTREE-INDEX.
 STYLE-WARNING: Implicitly creating new generic function POPULATE.

 Berkeley DB error #22: Invalid argument
    [Condition of type BDB-DB-ERROR]

Any help/tips/pointers very much appreciated.

Regards,

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sky at viridian-project.de  Fri Nov 13 10:02:28 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Fri, 13 Nov 2009 11:02:28 +0100 (CET)
Subject: [elephant-devel] "file is not of required architecture"
In-Reply-To: <hbsy7p7y.fsf@vps203.linuxvps.org>
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
	<24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>
	<3a4j7onz.fsf@vps203.linuxvps.org>
	<718A5688-F55B-49D6-8371-243B1358AF46@clozure.com>
	<y6mbprrk.fsf@vps203.linuxvps.org> <hbsy7p7y.fsf@vps203.linuxvps.org>
Message-ID: <9ab3e38b28bfdcd840af9091da075102.squirrel@mail.stardawn.org>


Sebastian Tennant wrote:
> Quoth Sebastian Tennant <sebyte at smolny.plus.com>:
>> Now let's see if I can get elephant's bdb store working on my Mac.
>
> Well, I'm a lot closer, but still not quite there yet.
>
> I compiled a 64-bit BerkeleyDB but now I get a 'Berkeley DB error #22: Invalid
> argument' error message when I try to (open-store ...):
>
>  bash-3.2$ cat elephant_fail.txt | grep -v '^\;\|^$'
>  Attempting to load libmemutil.dylib...
>  Loaded /Users/admin/cl/clbuild/source/elephant/src/memutil/libmemutil.dylib
>  /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c: In function
> ?lisp_compare2?:
>  /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c:1061: warning:
> unused variable ?i?
>  /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c: In function
> ?lisp_compare_key2?:
>  /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.c:1155: warning:
> unused variable ?i?
>  Loading /usr/local/BerkeleyDB.4.7/lib/libdb-4.7.dylib
>  Attempting to load libberkeley-db.dylib...
>  Loaded /Users/admin/cl/clbuild/source/elephant/src/db-bdb/libberkeley-db.dylib
>  STYLE-WARNING: Implicitly creating new generic function START-DEADLOCK-DETECTOR.
>  STYLE-WARNING: Implicitly creating new generic function STOP-DEADLOCK-DETECTOR.
>  STYLE-WARNING: Implicitly creating new generic function CHECKPOINT.
>  STYLE-WARNING: Implicitly creating new generic function BUILD-BTREE-INDEX.
>  STYLE-WARNING: Implicitly creating new generic function POPULATE.
>
>  Berkeley DB error #22: Invalid argument
>     [Condition of type BDB-DB-ERROR]
>
> Any help/tips/pointers very much appreciated.

Try again after removing libmemutil.so (thus causing Elephant to rebuild it).

Also check whether BDB now works correctly on your system. It should, but
let's make sure:

  http://www.oracle.com/technology/documentation/berkeley-db/db/programmer_reference/build_unix_test.html

  Leslie




From sebyte at smolny.plus.com  Fri Nov 13 10:18:49 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Fri, 13 Nov 2009 10:18:49 +0000
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian
	box	either
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<cd587bcc15f4024e7e9dd518f304abe4.squirrel@mail.stardawn.org>
Message-ID: <d43m7mnq.fsf@vps203.linuxvps.org>

Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
> I don't know much about this. My guess is that this page only lists
> authors that have committed a chunk big enough to be relevant in
> copyright matters.

Yes, I'm sure you're right.

> I've contributed a substantial number of fixes and help people when
> they face problems but I don't have the time to develop new features.

Well, as far as I'm concerned, you definitely deserve a mention.  Perhaps there
should be 'Contributors' section, or a 'Maintainers - Past & Present' section,
or both.

> Well there are more but I don't know what use they make of Elephant
> so I haven't mentioned them.

Fair enough.

> It's well known that Snow Leopard breaks a huge lot of MacPorts stuff,
> unfortunately (you can tell from that I'm not even a Mac user to begin
> with).

I was thinking of installing the Snow Leopard update but didn't get around to
it, and thank goodness I didn't!  My machine is an ordinary yellow and black
spotted Leopard.

> I agree that the second platform you've mentioned is pretty common but my
> statement was to be understood in a general case. I.e. we have to tend to a
> lot of software combinations -- this doesn't necessarily mean that it will
> automatically work on every one of those even if it's a common one.

Absolutely.

> Elephant does funky things like building a shared library in its build
> process, stressing the MOP heavily and using UFFI internals that don't cope
> well with the UFFI emulation of CFFI.

I thought CFFI emulated (or improved upon) UFFI somehow, not the other way
around.  [To be honest, I don't know the first thing about all this].

> It doesn't help of course that clbuild hasn't agreed to provide the real UFFI
> for Elephant so far.

Yes, this is another big stumbling block that really needs to be resolved.  My
cross-posted article to the CFFI list prompted Lu?s to reassert his willingness
to change the name of CFFI's uffi.asd once he received the necessary patches.
In his own words:

 "Patches for CFFI and, most importantly, clbuild are very much welcome. :-)"

The first thing I thought when I read this was:

 "Isn't Lu?s the author of CFFI?"

Anyway, it seems pretty clear to me that a coordinated joint effort is really
what's called for, involving at least one person from each interested party;
cffi, uffi, elephant, clbuild, and more.

What do you think?

> Thanks for being constructive. I agree that the next objective should be
> avoiding that new users fall into the same traps over and over.

Agreed.

> I already guessed that you were just pretty exhausted and disappointed
> after hours of struggling so I didn't really resent your complaints.

You guessed right. :)

> Sorry that it didn't work well for you right away.

It's not your fault Lelie, and there's therefore no need for you to apologise.
We'll get it working eventually and improve its usability in the process.

>>  1) If you use clbuild - edit wnpp-projects so that elephant-1.0 is pulled from
>>     darcs.
>>  2) Install UFFI (available from URL) and ensure that the 'real' uffi.asd is
>>     found by ASDF first.
>>  3) If you want to use a BerkeleyDB store, disregard the manual and install
>>     Berkeley DB version 4.7 (from URL).  Better still, update the manual.
>>  4) If you have a 64-bit machine make sure you compile a 64-bit BerkeleyDB.
>>     Elephant will expect it. Include instructions for said compilation.

> So does it work for you now or do you still face that obscure problem
> with bdb-constants.lisp which looks like that file was truncated on your
> installation?

Yes, I'm afraid I am.  The first time I try to call 'open-store', which cause
sbcl to hang, it is in the middle of doing some further compilation that begins
when the call to 'open-store' is evaluated.  The second time I try to call
'open-store' I get the truncated file error.  I've no doubt the file is
trunctaed because of the previous hang, i.e., the hang is the real error.

> This seems to be the right place as it's referenced from "Downloads":
>
>   http://common-lisp.net/project/elephant/doc/Installation.html#Installation

Presumably I'll need a common-lisp.net account in order to edit the HTML
source?  Shall I write and request one?

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From henrik at evahjelte.com  Fri Nov 13 11:20:17 2009
From: henrik at evahjelte.com (Henrik Hjelte)
Date: Fri, 13 Nov 2009 12:20:17 +0100
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian box
	either
In-Reply-To: <3a4jr6o8.fsf@vps203.linuxvps.org>
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
Message-ID: <50e8e4f60911130320l2a65c94dk11677012fab78536@mail.gmail.com>

On Thu, Nov 12, 2009 at 6:31 PM, Sebastian Tennant
<sebyte at smolny.plus.com> wrote:
>
> ?end of file on #<SB-SYS:FD-STREAM for "file
> ?/home/sebyte/lisp/cl/clbuild/source/elephant/src/db-bdb/berkeley-constants" {BB13669}>
> ? ?[Condition of type END-OF-FILE] ? ? ? ? ? ? ? ? ? ? ? ^^^^^^^^^^^^^^^^^^
>
> If I repeat it a thrid time I get the same error, so there's definitely a
> problem with berkeley-constants.
>
> For the last time:
>
> ?Debian Lenny, i386 32-bit
> ?SBCL 1.0.32.14
> ?elephant-1.0 from darcs
> ?BDB 4.7
> ?All other dependencies up-to-date as far as clbuild is concerned.

You could perhaps try to remove all fasl and generated files and
recompile with an older sbcl version. Sbcl can have bugs sometimes.

> I'm starting to wonder if many people really use elephant, and whether any
> elephant developers actually read this list?

I and Alex use elephant on GNU/Linux, but the postgresql backend.

> In short, I'm starting to become pretty disillusioned with elephant in
> particular and the state of Common Lisp libraries generally.
> Is it because I'm using bleeding edge versions of everything I wonder?

Yes, that would be my guess.

/Henrik



From rsynnott at gmail.com  Fri Nov 13 12:54:02 2009
From: rsynnott at gmail.com (Robert Synnott)
Date: Fri, 13 Nov 2009 12:54:02 +0000
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian box
	either
In-Reply-To: <ljia7rku.fsf@vps203.linuxvps.org>
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
Message-ID: <24f203480911130454p5846a223qb22599d2d27ff7f7@mail.gmail.com>

2009/11/13 Sebastian Tennant <sebyte at smolny.plus.com>:
>>> I'm starting to wonder if many people really use elephant,
>
>> Ian and me are using it every day in GNU/Linux environments..
>
> Two isn't many :)
>


There are certainly more than two. I've used it in a reasonably large
production webapp, and I know of a few other people who've used it.

It's a Common Lisp library, so it's not reasonable to expect thousands
of users, but it is certainly not un-used.
Rob

-- 
Robert Synnott
http://myblog.rsynnott.com
MSN: rsynnott at gmail.com
Jabber: rsynnott at gmail.com



From eslick at media.mit.edu  Fri Nov 13 18:19:54 2009
From: eslick at media.mit.edu (Ian Eslick)
Date: Fri, 13 Nov 2009 13:19:54 -0500
Subject: [elephant-devel] BDB problems and a proper FAQ
In-Reply-To: <ljia7rku.fsf@vps203.linuxvps.org>
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
Message-ID: <7212A9EA-9836-4714-81FE-B36AAD3FB83E@media.mit.edu>

Hi All,

To address recent discussions.

I'm ostensibly the primary developer on Elephant; the three prior  
primary developers have been inactive for years now.  Leslie has  
recently been kind enough to step up and help with user support and  
bug fixes and will certainly be recognized in the 1.0 manual when  
we're ready for a final release (updating the manual has been one of  
the gating tasks).  We all do our best to help and want Elephant to be  
as robust as possible.  I in particular have been lousy at responding  
recently, but there are of course good reasons for that.

Anyway, I'm sorry to hear that your experience has been so  
frustrating.  Unfortunately, the 64-bit lisp + OS + BDB combinations  
have made things much more difficult than they once were.  As an  
aside, Franz AllegroCache has most of the features of Elephant and has  
a great and responsive support staff.  If you are frustrated with the  
state of free, volunteer-developed lisp libraries it is possible to  
pay to get more reliability and better support!

There are dozens of users of Elephant, some casual and some serious.   
Some are working on older versions of the library as their deployed  
systems are stable and not actively developed so not everyone is  
subject to problems with the current release.  Leslie and I are just  
the most vocal on this list.  I'm running Elephant under ClozureCL  
these days and it works fine under Snow Leopard as well as Debian.

What's your current status, are you still stuck with the 'invalid  
argument' problems?


I think that a FAQ on the home page would be a great way to keep track  
of the kind of issues that you have experienced.  Here is a first  
crack at it; please add your own experiences/observations and I'll  
post it on the home page.

1) Documentation for 1.0 release

The documentation on the website is from the 0.9.1 release and may be  
incorrect in some details.  The mailing list since 2008 and this FAQ  
should supersede anything in the documentation.

2) CLBUILD Compatability

Elephant does not work properly under the default cl-build  
configuration due to it's UFFI dependency.  We have made attempts to  
port BDB to CFFI without success;  there are subtle issues that lead  
to obscure bugs and the original FFI developer who wrote that code  
left the project in 2003 or 2004 and no one else has yet had the time  
and expertise to resolve them.  The workaround, thanks to Sebastian  
Tennant, is:

a) Edit wnpp-projects so that elephant-1.0 is pulled from darcs.
b) Install UFFI (available from URL) and ensure that the 'real'  
uffi.asd is found by ASDF first.

3) Library and Lisp Dependencies

The most common active configurations of the 1.0 release branch is on  
SBCL/Linux/Postmodern+BDB and ClozureCL/MacOS/BDB.  Elephant is known  
to work on those platforms when properly configured.

We track the latest version of BDB.  The current 1.0 pre-release  
depends on BDB 4.7.

4) 64-bit Lisp and BDB

If you are running a 32-bit or 64-bit lisp, BDB will need to be built  
for the appropriate architecture.  Due to this the default MacPorts,  
Debian packages or  other pre-packaged distributions of BDB may fail  
for you.  Building from source is the appropriate solution (thanks to  
Matthew Emerson for the script).  Don't forget to update my- 
config.sexp to use "4.7" and point to the appropriate paths.

a. get bdb 4.7

Direct link; if it doesn't work use the index page:
http://download.oracle.com/berkeley-db/db-4.7.25.tar.gz

Index page:
http://www.oracle.com/technology/software/products/berkeley-db/db/index.html

b. unpack into /usr/local/src (or wherever)

Build docs are at

file:///usr/local/src/db-4.7.25/docs/ref/build_unix/intro.html

(or wherever you put it)

c. cd to db-4.7.25/build_unix

d. use the following command line to configure, where $ is the shell  
prompt

$ CFLAGS="-arch x86_64" LDFLAGS="-arch x86_64" ../dist/configure

e. sudo make install


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/elephant-devel/attachments/20091113/a2b42f22/attachment.html>

From sebyte at smolny.plus.com  Fri Nov 13 18:20:43 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Fri, 13 Nov 2009 18:20:43 +0000
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian box
	either
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<24f203480911130454p5846a223qb22599d2d27ff7f7@mail.gmail.com>
Message-ID: <d43mjngk.fsf@vps203.linuxvps.org>

Quoth Robert Synnott <rsynnott at gmail.com>:
> 2009/11/13 Sebastian Tennant <sebyte at smolny.plus.com>:
>> Two isn't many :)

> There are certainly more than two. I've used it in a reasonably large
> production webapp, and I know of a few other people who've used it.

I was joking (hence the smiley) and trying to express a little of my
exasperation at the same time.  I didn't think for one minute there were only
two people in the world using elephant!  Still, it's good to hear that you, and
others, are using it too.

> It's a Common Lisp library, so it's not reasonable to expect thousands of
> users,

Sadly, or perhaps not so, you are right.

> but it is certainly not un-used.

It's unused by me!  I can't get the damn thing to work! :)

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sky at viridian-project.de  Fri Nov 13 19:17:18 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Fri, 13 Nov 2009 20:17:18 +0100 (CET)
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian
 box	either
In-Reply-To: <d43m7mnq.fsf@vps203.linuxvps.org>
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<cd587bcc15f4024e7e9dd518f304abe4.squirrel@mail.stardawn.org>
	<d43m7mnq.fsf@vps203.linuxvps.org>
Message-ID: <e3e8a7a46e511f2eafcf8d11255622a1.squirrel@mail.stardawn.org>


Sebastian Tennant wrote:

> I thought CFFI emulated (or improved upon) UFFI somehow, not the other way
> around.  [To be honest, I don't know the first thing about all this].

CFFI is younger than UFFI and works in a different way. They provide an
incomplete UFFI compatibility layer, though.


>  "Patches for CFFI and, most importantly, clbuild are very much welcome. :-)"
>
> The first thing I thought when I read this was:
>
>  "Isn't Lu?s the author of CFFI?"

He's one of the primary developers I think. But that doesn't mean that
he's able or willing to do this work. Remember this is open source so
people often need to provide ready-to-commit patches to get their stuff in.


> Anyway, it seems pretty clear to me that a coordinated joint effort is really
> what's called for, involving at least one person from each interested party;
> cffi, uffi, elephant, clbuild, and more.

Why wouldn't one person be enough, provided they understand the matter well
enough?


> Yes, I'm afraid I am.  The first time I try to call 'open-store', which cause
> sbcl to hang, it is in the middle of doing some further compilation that begins
> when the call to 'open-store' is evaluated.

No I think this particular compilation has already been completed at this point.


> The second time I try to call
> 'open-store' I get the truncated file error.  I've no doubt the file is
> trunctaed because of the previous hang, i.e., the hang is the real error.

Not sure about that. We need a backtrace, and please show us your
bdb-constants.lisp.


> Presumably I'll need a common-lisp.net account in order to edit the HTML
> source?  Shall I write and request one?

Yes. Request a plain account without any group access and get the files
from /project/elephant/public_html. Then make the changes and send a patch.

  Thanks!

    Leslie

-- 
http://www.linkedin.com/in/polzer




From sky at viridian-project.de  Fri Nov 13 19:19:19 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Fri, 13 Nov 2009 20:19:19 +0100 (CET)
Subject: [elephant-devel] BDB problems and a proper FAQ
In-Reply-To: <7212A9EA-9836-4714-81FE-B36AAD3FB83E@media.mit.edu>
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<7212A9EA-9836-4714-81FE-B36AAD3FB83E@media.mit.edu>
Message-ID: <ccfe5b2af0d4dfb6fa9b0b22f246fd65.squirrel@mail.stardawn.org>


Ian Eslick wrote:

> We track the latest version of BDB.  The current 1.0 pre-release
> depends on BDB 4.7.

Okay, but the latest is 4.8 since about two weeks so we might want
to mention that.


> a. get bdb 4.7
>
> Direct link; if it doesn't work use the index page:
> http://download.oracle.com/berkeley-db/db-4.7.25.tar.gz

Best to download patches too. 4.7.25.4 is the actual latest of the
4.7 branch.

  Leslie

-- 
http://www.linkedin.com/in/polzer




From sebyte at smolny.plus.com  Sat Nov 14 17:32:37 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Sat, 14 Nov 2009 17:32:37 +0000
Subject: [elephant-devel] "file is not of required architecture"
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
	<24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>
	<3a4j7onz.fsf@vps203.linuxvps.org>
	<718A5688-F55B-49D6-8371-243B1358AF46@clozure.com>
	<y6mbprrk.fsf@vps203.linuxvps.org>
	<hbsy7p7y.fsf@vps203.linuxvps.org>
	<9ab3e38b28bfdcd840af9091da075102.squirrel@mail.stardawn.org>
Message-ID: <vdhdxb9m.fsf@vps203.linuxvps.org>

Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
>>  Berkeley DB error #22: Invalid argument
>>     [Condition of type BDB-DB-ERROR]

> Try again after removing libmemutil.so (thus causing Elephant to rebuild it).

It fails before this file is built, i.e., there is no libmemutil.so.  Instead,
there's a strange directory called libmemutil.dylib.dSYM, which is presumably a
temporary directory left behind because of the hang

> Also check whether BDB now works correctly on your system. It should, but
> let's make sure:
>
>   http://www.oracle.com/technology/documentation/berkeley-db/db/programmer_reference/build_unix_test.html

Noted. I'll do this as a last resort.

First of all, I'm going to go back to a stable release of SBCL, rebuild bdb 4.7
with the four patches you mention applied and take it from there.

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sebyte at smolny.plus.com  Sat Nov 14 17:57:22 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Sat, 14 Nov 2009 17:57:22 +0000
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian
	box	either
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<cd587bcc15f4024e7e9dd518f304abe4.squirrel@mail.stardawn.org>
	<d43m7mnq.fsf@vps203.linuxvps.org>
	<e3e8a7a46e511f2eafcf8d11255622a1.squirrel@mail.stardawn.org>
Message-ID: <r5s1xa4d.fsf@vps203.linuxvps.org>

Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
> CFFI is younger than UFFI and works in a different way. They provide an
> incomplete UFFI compatibility layer, though.

Thanks for the clarification.

> [Lu?s is] one of the primary [CFFI] developers I think. But that doesn't mean
> that he's able [...] to do this work.

If he isn't able to do it, then who is?

> Why wouldn't one person be enough, provided they understand the matter well
> enough?

Does such a person exist?

> No I think this particular compilation has already been completed at this point.
> [...]
> Not sure about that. We need a backtrace, and please show us your
> bdb-constants.lisp.

I have a third machine (an x86 32-bit Debian VPS) and I've finally got
elephant-1.0 to work with a patched bdb 4.7 and SBCL 1.0.30.

I've been using git SBCL (1.0.32.14) until now so I'm beginning to suspect SBCL
(as Henrik also suggested).

My next move will be to go back to the latest SBCL release tarball and take it
from there.

> Yes. Request a plain account without any group access and get the files
> from /project/elephant/public_html. Then make the changes and send a patch.

It seems this is no longer necessary now that Ian's on the case.  I still
intend to provide some patches for doc/Makefile and doc/elephant.texinfo.

Many thanks for all your help.

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sebyte at smolny.plus.com  Sat Nov 14 18:21:53 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Sat, 14 Nov 2009 18:21:53 +0000
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian box
	either
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<50e8e4f60911130320l2a65c94dk11677012fab78536@mail.gmail.com>
Message-ID: <my2px8zi.fsf@vps203.linuxvps.org>

Quoth Henrik Hjelte <henrik at evahjelte.com>:
> You could perhaps try to remove all fasl and generated files 

I've been doing that all along.

> and recompile with an older sbcl version. Sbcl can have bugs sometimes.

This is what I'm going to do.  Thanks for the suggestion.  I despaired at the
thought initially but then I remembered that I have SBCL 1.0.30 installed on my
32-bit x86 Debian VPS and sure enough, I've finally got elephant to work over
there.

>> I'm starting to wonder if many people really use elephant, and whether any
>> elephant developers actually read this list?
>
> I and Alex use elephant on GNU/Linux, but the postgresql backend.

That's still good to know.  My frustration got the better of me :)

>> Is it because I'm using bleeding edge versions of everything I wonder?
>
> Yes, that would be my guess.

And it looks like you might be right (with respect to SBCL).

Thanks.

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From eslick at media.mit.edu  Sat Nov 14 21:11:18 2009
From: eslick at media.mit.edu (Ian Eslick)
Date: Sat, 14 Nov 2009 16:11:18 -0500
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian
	box	either
In-Reply-To: <r5s1xa4d.fsf@vps203.linuxvps.org>
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<cd587bcc15f4024e7e9dd518f304abe4.squirrel@mail.stardawn.org>
	<d43m7mnq.fsf@vps203.linuxvps.org>
	<e3e8a7a46e511f2eafcf8d11255622a1.squirrel@mail.stardawn.org>
	<r5s1xa4d.fsf@vps203.linuxvps.org>
Message-ID: <F4ABC42E-9337-429C-A758-D9C91408A90C@media.mit.edu>

>> No I think this particular compilation has already been completed  
>> at this point.
>> [...]
>> Not sure about that. We need a backtrace, and please show us your
>> bdb-constants.lisp.
>
> I have a third machine (an x86 32-bit Debian VPS) and I've finally got
> elephant-1.0 to work with a patched bdb 4.7 and SBCL 1.0.30.


Do you think this explains your problems on other platforms?  BTW -  
I'd recommend using Clozure CL on Mac instead of SBCL (I'm not sure if  
SBCL threading on Mac OS is reliable yet, it wasn't when I switched 9  
months or so ago).  There are tradeoffs between the two, but I've  
found working with ccl to be a better experience overall.

>> Yes. Request a plain account without any group access and get the  
>> files
>> from /project/elephant/public_html. Then make the changes and send  
>> a patch.
>
> It seems this is no longer necessary now that Ian's on the case.  I  
> still
> intend to provide some patches for doc/Makefile and doc/ 
> elephant.texinfo.
>

I'm happy to make the updates if you provide doc patches (for the 1.0  
release) and I'll post the FAQ once we've resolved the current problems.

Thanks,
Ian



From sky at viridian-project.de  Sun Nov 15 09:20:10 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Sun, 15 Nov 2009 10:20:10 +0100 (CET)
Subject: [elephant-devel] "file is not of required architecture"
In-Reply-To: <vdhdxb9m.fsf@vps203.linuxvps.org>
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
	<24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>
	<3a4j7onz.fsf@vps203.linuxvps.org>
	<718A5688-F55B-49D6-8371-243B1358AF46@clozure.com>
	<y6mbprrk.fsf@vps203.linuxvps.org> <hbsy7p7y.fsf@vps203.linuxvps.org>
	<9ab3e38b28bfdcd840af9091da075102.squirrel@mail.stardawn.org>
	<vdhdxb9m.fsf@vps203.linuxvps.org>
Message-ID: <2192b07c4a8ef74499a6de2ec4b0a89e.squirrel@mail.stardawn.org>


Sebastian Tennant wrote:
> Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
>>>  Berkeley DB error #22: Invalid argument
>>>     [Condition of type BDB-DB-ERROR]
>
>> Try again after removing libmemutil.so (thus causing Elephant to rebuild it).
>
> It fails before this file is built, i.e., there is no libmemutil.so.  Instead,
> there's a strange directory called libmemutil.dylib.dSYM, which is presumably a
> temporary directory left behind because of the hang

Can you send us the backtrace?

Also try compiling libmemutil.so manually.


> First of all, I'm going to go back to a stable release of SBCL, rebuild bdb 4.7
> with the four patches you mention applied and take it from there.

The patches shouldn't influence the compatibility but it can't harm to
apply them either.

  Leslie

-- 
http://www.linkedin.com/in/polzer




From sky at viridian-project.de  Sun Nov 15 09:24:40 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Sun, 15 Nov 2009 10:24:40 +0100 (CET)
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian
 box	either
In-Reply-To: <r5s1xa4d.fsf@vps203.linuxvps.org>
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<cd587bcc15f4024e7e9dd518f304abe4.squirrel@mail.stardawn.org>
	<d43m7mnq.fsf@vps203.linuxvps.org>
	<e3e8a7a46e511f2eafcf8d11255622a1.squirrel@mail.stardawn.org>
	<r5s1xa4d.fsf@vps203.linuxvps.org>
Message-ID: <6a1e510cdd648a923e8edacc787a8081.squirrel@mail.stardawn.org>


Sebastian Tennant wrote:

>> [Lu?s is] one of the primary [CFFI] developers I think. But that doesn't mean
>> that he's able [...] to do this work.
>
> If he isn't able to do it, then who is?

Well, with "able" I mostly meant "has the time, motivation and energy".


>> Why wouldn't one person be enough, provided they understand the matter well
>> enough?
>
> Does such a person exist?

I think there are a few candidates that could fix either Elephant or CFFI-UFFI
(given enough time) and a whole host (including you) that could provide clbuild
patches or a custom Elephant install script.


> It seems this is no longer necessary now that Ian's on the case.

Ian's a busy man though so please consider helping him out nevertheless.


> I still intend to provide some patches for doc/Makefile and doc/elephant.texinfo.

Thank you! And


> Many thanks for all your help.

you're welcome, too.

  Leslie

-- 
http://www.linkedin.com/in/polzer




From sebyte at smolny.plus.com  Thu Nov 19 19:10:18 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Thu, 19 Nov 2009 19:10:18 +0000
Subject: [elephant-devel] BDB problems and a proper FAQ
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<7212A9EA-9836-4714-81FE-B36AAD3FB83E@media.mit.edu>
Message-ID: <k4xm49gl.fsf@vps203.linuxvps.org>

Hi Ian,

Sorry for the late reply.  I have a rather annoying cold at the moment.

Quoth Ian Eslick <eslick at media.mit.edu>:
> I'm ostensibly the primary developer on Elephant; the three prior primary
> developers have been inactive for years now.  Leslie has recently been kind
> enough to step up and help with user support and bug fixes and will certainly
> be recognized in the 1.0 manual when we're ready for a final release
> (updating the manual has been one of the gating tasks).  We all do our best
> to help and want Elephant to be as robust as possible.  I in particular have
> been lousy at responding recently, but there are of course good reasons for
> that.

I've no doubt there are.  Glad to hear that Leslie will be getting a mention in
the 1.0 manual.  He's been very patient with me!!!

> Anyway, I'm sorry to hear that your experience has been so frustrating.
> Unfortunately, the 64-bit lisp + OS + BDB combinations have made things much
> more difficult than they once were.  As an aside, Franz AllegroCache has most
> of the features of Elephant and has a great and responsive support staff.  If
> you are frustrated with the state of free, volunteer-developed lisp libraries
> it is possible to pay to get more reliability and better support!

Noted.

> There are dozens of users of Elephant, some casual and some serious.  Some
> are working on older versions of the library as their deployed systems are
> stable and not actively developed so not everyone is subject to problems with
> the current release.  

Naturally.

> Leslie and I are just the most vocal on this list.

Understood.

> I'm running Elephant under ClozureCL these days and it works fine under Snow
> Leopard as well as Debian.

Noted.

> What's your current status, are you still stuck with the 'invalid argument'
> problems?

Yeah, I'm afraid so (on my Mac).  I've got an elephant-1.0 bdb 4.7 store
working on a 32-bit Debian box with SBCL 1.0.30 but am having no luck with the
Mac, even with CCL.  I'll post a backtrace in a separate thread.

> I think that a FAQ on the home page would be a great way to keep track of the
> kind of issues that you have experienced.  Here is a first crack at it;
> please add your own experiences/observations and I'll post it on the home
> page.

You've covered pretty much everything.  Please find attached HTML (which I hope
is a help).

Regards,

Sebastian
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/elephant-devel/attachments/20091119/88b33510/attachment.html>

From eslick at media.mit.edu  Thu Nov 19 19:58:56 2009
From: eslick at media.mit.edu (Ian Eslick)
Date: Thu, 19 Nov 2009 11:58:56 -0800
Subject: [elephant-devel] BDB problems and a proper FAQ
In-Reply-To: <k4xm49gl.fsf@vps203.linuxvps.org>
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<7212A9EA-9836-4714-81FE-B36AAD3FB83E@media.mit.edu>
	<k4xm49gl.fsf@vps203.linuxvps.org>
Message-ID: <5A83734B-89A2-4E73-9440-B80A618BD2BF@media.mit.edu>

What's your my-config look like again?

Sent from my iPhone

On Nov 19, 2009, at 11:10 AM, Sebastian Tennant  
<sebyte at smolny.plus.com> wrote:

> Hi Ian,
>
> Sorry for the late reply.  I have a rather annoying cold at the  
> moment.
>
> Quoth Ian Eslick <eslick at media.mit.edu>:
>> I'm ostensibly the primary developer on Elephant; the three prior  
>> primary
>> developers have been inactive for years now.  Leslie has recently  
>> been kind
>> enough to step up and help with user support and bug fixes and will  
>> certainly
>> be recognized in the 1.0 manual when we're ready for a final release
>> (updating the manual has been one of the gating tasks).  We all do  
>> our best
>> to help and want Elephant to be as robust as possible.  I in  
>> particular have
>> been lousy at responding recently, but there are of course good  
>> reasons for
>> that.
>
> I've no doubt there are.  Glad to hear that Leslie will be getting a  
> mention in
> the 1.0 manual.  He's been very patient with me!!!
>
>> Anyway, I'm sorry to hear that your experience has been so  
>> frustrating.
>> Unfortunately, the 64-bit lisp + OS + BDB combinations have made  
>> things much
>> more difficult than they once were.  As an aside, Franz  
>> AllegroCache has most
>> of the features of Elephant and has a great and responsive support  
>> staff.  If
>> you are frustrated with the state of free, volunteer-developed lisp  
>> libraries
>> it is possible to pay to get more reliability and better support!
>
> Noted.
>
>> There are dozens of users of Elephant, some casual and some  
>> serious.  Some
>> are working on older versions of the library as their deployed  
>> systems are
>> stable and not actively developed so not everyone is subject to  
>> problems with
>> the current release.
>
> Naturally.
>
>> Leslie and I are just the most vocal on this list.
>
> Understood.
>
>> I'm running Elephant under ClozureCL these days and it works fine  
>> under Snow
>> Leopard as well as Debian.
>
> Noted.
>
>> What's your current status, are you still stuck with the 'invalid  
>> argument'
>> problems?
>
> Yeah, I'm afraid so (on my Mac).  I've got an elephant-1.0 bdb 4.7  
> store
> working on a 32-bit Debian box with SBCL 1.0.30 but am having no  
> luck with the
> Mac, even with CCL.  I'll post a backtrace in a separate thread.
>
>> I think that a FAQ on the home page would be a great way to keep  
>> track of the
>> kind of issues that you have experienced.  Here is a first crack at  
>> it;
>> please add your own experiences/observations and I'll post it on  
>> the home
>> page.
>
> You've covered pretty much everything.  Please find attached HTML  
> (which I hope
> is a help).
>
> Regards,
>
> Sebastian
> -- 
> Emacs' AlsaPlayer - Music Without Jolts
> Lightweight, full-featured and mindful of your idyllic happiness.
> http://home.gna.org/eap
>
> <elephant-1.0.html>
> _______________________________________________
> elephant-devel site list
> elephant-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/elephant-devel



From eslick at media.mit.edu  Thu Nov 19 21:37:44 2009
From: eslick at media.mit.edu (Ian Eslick)
Date: Thu, 19 Nov 2009 13:37:44 -0800
Subject: [elephant-devel] "file is not of required architecture"
In-Reply-To: <hbt28byd.fsf@vps203.linuxvps.org>
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
Message-ID: <E6814B9C-771C-4D7A-828A-2B5ECEDAC921@media.mit.edu>

Seb,

The reason I asked for the my-config.sexp is that the my-config you  
forwarded some time ago below will fail in the way you are talking  
about (undefined symbols as well as unrecognized arguments) because  
this file is telling it to use BDB 4.5 headers to talk to 4.7  
libraries.  BDB changes the constant mapping with every .X increment  
requiring some manual work and a condition to test for which version  
of the constants to use.

Ian

On Nov 10, 2009, at 10:35 AM, Sebastian Tennant wrote:

> my-config.sexp now reads:
>
> ((:compiler . :gcc)
>  (:berkeley-db-version . "4.5")
>  (:berkeley-db-include-dir . "/opt/local/include/db47/")
>  (:berkeley-db-lib-dir . "/opt/local/lib/db47/")
>  (:berkeley-db-lib . "/opt/local/lib/db47/libdb-4.7.dylib")
>  (:berkeley-db-deadlock . "/opt/local/bin/db47_deadlock")
>  (:berkeley-db-cachesize . 20971520)
>  (:berkeley-db-max-locks . 2000)
>  (:berkeley-db-max-objects . 2000)
>  (:berkeley-db-map-degree2 . t)
>  (:berkeley-db-mvcc . nil)
>  (:clsql-lib-paths . nil)
>  (:prebuilt-libraries . nil))




From sebyte at smolny.plus.com  Thu Nov 19 22:59:37 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Thu, 19 Nov 2009 22:59:37 +0000
Subject: [elephant-devel] BDB problems and a proper FAQ
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<7212A9EA-9836-4714-81FE-B36AAD3FB83E@media.mit.edu>
	<k4xm49gl.fsf@vps203.linuxvps.org>
	<5A83734B-89A2-4E73-9440-B80A618BD2BF@media.mit.edu>
Message-ID: <d43e3yue.fsf@vps203.linuxvps.org>

Quoth Ian Eslick <eslick at media.mit.edu>:
> What's your my-config look like again?

I removed some stale .dx32fsl files and I'm pleased to say that bdb stores are
now working with a 32-bit CCL on my Mac.  (I ended up in '32-bit mode' while
wrangling with SBCL, which I still haven't managed to get working).

I'm going to leave it at that for now, in terms of testing various
configurations and actually start using it a little :)

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sebyte at smolny.plus.com  Thu Nov 19 23:09:03 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Thu, 19 Nov 2009 23:09:03 +0000
Subject: [elephant-devel] "file is not of required architecture"
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<E6814B9C-771C-4D7A-828A-2B5ECEDAC921@media.mit.edu>
Message-ID: <8we23yeo.fsf@vps203.linuxvps.org>

Quoth Ian Eslick <eslick at media.mit.edu>:
> The reason I asked for the my-config.sexp is that the my-config you  
> forwarded some time ago below will fail in the way you are talking  
> about (undefined symbols as well as unrecognized arguments) because  
> this file is telling it to use BDB 4.5 headers to talk to 4.7  
> libraries.  BDB changes the constant mapping with every .X increment  
> requiring some manual work and a condition to test for which version  
> of the constants to use.

Yes, I did an 'M-x replace-string 45 RET 47 RET' on config.sexp and lived with
the results for a while before picking up on that.  (I think the gods were
conspiring against me for a while back there!)

Still, I hope you agree that some good has come of it all.  Thanks for your
support in any case.

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From eslick at media.mit.edu  Thu Nov 19 23:27:58 2009
From: eslick at media.mit.edu (Ian Eslick)
Date: Thu, 19 Nov 2009 15:27:58 -0800
Subject: [elephant-devel] "file is not of required architecture"
In-Reply-To: <8we23yeo.fsf@vps203.linuxvps.org>
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<E6814B9C-771C-4D7A-828A-2B5ECEDAC921@media.mit.edu>
	<8we23yeo.fsf@vps203.linuxvps.org>
Message-ID: <EAE96497-3DE0-482E-8619-FC33DABF688D@media.mit.edu>

No problem!  I'll add the FAQ to the site shortly.  -Ian

On Nov 19, 2009, at 3:09 PM, Sebastian Tennant wrote:

> Quoth Ian Eslick <eslick at media.mit.edu>:
>> The reason I asked for the my-config.sexp is that the my-config you
>> forwarded some time ago below will fail in the way you are talking
>> about (undefined symbols as well as unrecognized arguments) because
>> this file is telling it to use BDB 4.5 headers to talk to 4.7
>> libraries.  BDB changes the constant mapping with every .X increment
>> requiring some manual work and a condition to test for which version
>> of the constants to use.
>
> Yes, I did an 'M-x replace-string 45 RET 47 RET' on config.sexp and  
> lived with
> the results for a while before picking up on that.  (I think the  
> gods were
> conspiring against me for a while back there!)
>
> Still, I hope you agree that some good has come of it all.  Thanks  
> for your
> support in any case.
>
> Seb
> -- 
> Emacs' AlsaPlayer - Music Without Jolts
> Lightweight, full-featured and mindful of your idyllic happiness.
> http://home.gna.org/eap
>
>
> _______________________________________________
> elephant-devel site list
> elephant-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/elephant-devel




From sebyte at smolny.plus.com  Sat Nov 21 21:54:42 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Sat, 21 Nov 2009 21:54:42 +0000
Subject: [elephant-devel] "file is not of required architecture"
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
	<24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>
	<3a4j7onz.fsf@vps203.linuxvps.org>
	<718A5688-F55B-49D6-8371-243B1358AF46@clozure.com>
	<y6mbprrk.fsf@vps203.linuxvps.org>
	<hbsy7p7y.fsf@vps203.linuxvps.org>
	<9ab3e38b28bfdcd840af9091da075102.squirrel@mail.stardawn.org>
	<vdhdxb9m.fsf@vps203.linuxvps.org>
	<2192b07c4a8ef74499a6de2ec4b0a89e.squirrel@mail.stardawn.org>
Message-ID: <skc71r31.fsf@vps203.linuxvps.org>

Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
> Can you send us the backtrace?  Also try compiling libmemutil.so manually.
> The patches shouldn't influence the compatibility but it can't harm to apply
> them either.

I've got elephant working on my Mac as well now - a 32-bit ClozureCL and a
patched 32-bit BDB - so that's it for now :) No more configuration testing.
Thanks again for your help.

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From eslick at media.mit.edu  Sat Nov 21 22:08:52 2009
From: eslick at media.mit.edu (Ian Eslick)
Date: Sat, 21 Nov 2009 14:08:52 -0800
Subject: [elephant-devel] "file is not of required architecture"
In-Reply-To: <skc71r31.fsf@vps203.linuxvps.org>
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
	<24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>
	<3a4j7onz.fsf@vps203.linuxvps.org>
	<718A5688-F55B-49D6-8371-243B1358AF46@clozure.com>
	<y6mbprrk.fsf@vps203.linuxvps.org>
	<hbsy7p7y.fsf@vps203.linuxvps.org>
	<9ab3e38b28bfdcd840af9091da075102.squirrel@mail.stardawn.org>
	<vdhdxb9m.fsf@vps203.linuxvps.org>
	<2192b07c4a8ef74499a6de2ec4b0a89e.squirrel@mail.stardawn.org>
	<skc71r31.fsf@vps203.linuxvps.org>
Message-ID: <E8CD1C43-999C-4262-BA18-D9F8A1E0FC28@media.mit.edu>

What were the problems you encountered and solutions you found?   
Anything we should add to the FAQ?

Sent from my iPhone

On Nov 21, 2009, at 1:54 PM, Sebastian Tennant  
<sebyte at smolny.plus.com> wrote:

> Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
>> Can you send us the backtrace?  Also try compiling libmemutil.so  
>> manually.
>> The patches shouldn't influence the compatibility but it can't harm  
>> to apply
>> them either.
>
> I've got elephant working on my Mac as well now - a 32-bit ClozureCL  
> and a
> patched 32-bit BDB - so that's it for now :) No more configuration  
> testing.
> Thanks again for your help.
>
> Seb
> -- 
> Emacs' AlsaPlayer - Music Without Jolts
> Lightweight, full-featured and mindful of your idyllic happiness.
> http://home.gna.org/eap
>
>
> _______________________________________________
> elephant-devel site list
> elephant-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/elephant-devel



From sebyte at smolny.plus.com  Sat Nov 21 22:10:53 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Sat, 21 Nov 2009 22:10:53 +0000
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian
	box	either
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<cd587bcc15f4024e7e9dd518f304abe4.squirrel@mail.stardawn.org>
	<d43m7mnq.fsf@vps203.linuxvps.org>
	<e3e8a7a46e511f2eafcf8d11255622a1.squirrel@mail.stardawn.org>
	<r5s1xa4d.fsf@vps203.linuxvps.org>
	<6a1e510cdd648a923e8edacc787a8081.squirrel@mail.stardawn.org>
Message-ID: <ocmv1qc2.fsf@vps203.linuxvps.org>

Quoth "Leslie P. Polzer" <sky at viridian-project.de>:
> Sebastian Tennant wrote:
>> If he isn't able to do it, then who is?

> Well, with "able" I mostly meant "has the time, motivation and energy".

>> Does such a person exist?

> I think there are a few candidates that could fix either Elephant or
> CFFI-UFFI (given enough time) and a whole host (including you) that could
> provide clbuild patches or a custom Elephant install script.

The file 'wnpp-projects' just needs to be altered so that it pulls in version
1.0 (which I've already requested[1]) and it might be a good idea to remove
CFFI as a dependency so that users are presented with a "system not found"
error (rather than a more obscure compilation error) when they try to compile
elephant without UFFI installed.  Again, this is a case of simply altering the
elephant entry in the dependencies file.  I can request this as well if you
like.

Beyond that, I don't see how else clbuild can be 'patched'.

>> It seems this is no longer necessary now that Ian's on the case.

> Ian's a busy man though so please consider helping him out nevertheless.

Done.

Seb

[1] http://article.gmane.org/gmane.lisp.clbuild.devel/286

-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap





From sebyte at smolny.plus.com  Sat Nov 21 22:21:15 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Sat, 21 Nov 2009 22:21:15 +0000
Subject: [elephant-devel] "file is not of required architecture"
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
	<24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>
	<3a4j7onz.fsf@vps203.linuxvps.org>
	<718A5688-F55B-49D6-8371-243B1358AF46@clozure.com>
	<y6mbprrk.fsf@vps203.linuxvps.org>
	<hbsy7p7y.fsf@vps203.linuxvps.org>
	<9ab3e38b28bfdcd840af9091da075102.squirrel@mail.stardawn.org>
	<vdhdxb9m.fsf@vps203.linuxvps.org>
	<2192b07c4a8ef74499a6de2ec4b0a89e.squirrel@mail.stardawn.org>
	<skc71r31.fsf@vps203.linuxvps.org>
	<E8CD1C43-999C-4262-BA18-D9F8A1E0FC28@media.mit.edu>
Message-ID: <k4xj1pus.fsf@vps203.linuxvps.org>

Quoth Ian Eslick <eslick at media.mit.edu>:
> What were the problems you encountered and solutions you found?   
> Anything we should add to the FAQ?

Nothing (reliable) that isn't already addressed in the FAQ.  I suspect there's
something in git SBCL (1.0.32.xx) which was causing me problems on my 32-bit
Debian box but I'll have to go back to that machine and check again before I
can confirm this.

When (approximately) do you think you might like to formally release version
1.0?

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sebyte at smolny.plus.com  Sat Nov 21 22:35:03 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Sat, 21 Nov 2009 22:35:03 +0000
Subject: [elephant-devel] BDB stores don't work on my 32-bit Debian box
	either
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<cd587bcc15f4024e7e9dd518f304abe4.squirrel@mail.stardawn.org>
	<d43m7mnq.fsf@vps203.linuxvps.org>
	<e3e8a7a46e511f2eafcf8d11255622a1.squirrel@mail.stardawn.org>
	<r5s1xa4d.fsf@vps203.linuxvps.org>
	<F4ABC42E-9337-429C-A758-D9C91408A90C@media.mit.edu>
Message-ID: <fx871p7s.fsf_-_@vps203.linuxvps.org>

Quoth Ian Eslick <eslick at media.mit.edu>:
>> I still intend to provide some patches for doc/Makefile and doc/
>> elephant.texinfo.

> I'm happy to make the updates if you provide doc patches (for the 1.0  
> release) and I'll post the FAQ once we've resolved the current problems.

I hope to be able to edit elephant.texinfo so that we can add an 'info' target
to the Makefile _without_ breaking the existing targets; 'docs' (HTML) and
'pdf', but... I'm not a texinfo expert so no guarantees I'm afraid.

I'll also make minor changes to content where it obviously needs it (such as
"Elephant only works with Berkeley DB version 4.5").

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sebyte at smolny.plus.com  Sat Nov 21 23:50:33 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Sat, 21 Nov 2009 23:50:33 +0000
Subject: [elephant-devel] Inspecting a bdb store
Message-ID: <bpiv1lpy.fsf@vps203.linuxvps.org>

Hi guys,

I'm _very_ new to the Berkeley DB way of doing things so please bear with me.

map-root allows you to inspect the root BTree of a bdb store, but having
created a few instances of a persistent class and performed a map-root, there's
no sign of these stored instances.  get-instances-by-class retrieves them but
what if I didn't know the name of their class?

To put it another way, if someone were to give you a bdb folder/store that you
(or they) knew nothing about, how would you go about inspecting its contents?

Is it the 'class root' that I need to map somehow? [1]

Seb


[1] The manual says that the 'class root' will be discussed later but the
    two-word phrase 'class root' is never seen again.

-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From sebyte at smolny.plus.com  Sun Nov 22 13:23:43 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Sun, 22 Nov 2009 13:23:43 +0000
Subject: [elephant-devel] "file is not of required architecture"
References: <fx8ntlhu.fsf@vps203.linuxvps.org>
	<96737bf1af25a31816292570b4e95c2a.squirrel@mail.stardawn.org>
	<my2uag66.fsf@vps203.linuxvps.org>
	<20091110104433.GA19646@viridian-project.de>
	<hbt28byd.fsf@vps203.linuxvps.org>
	<20091110191756.GA6201@viridian-project.de>
	<7htv7t26.fsf@vps203.linuxvps.org>
	<24f203480911120551o409c3114l293b961fd25693e2@mail.gmail.com>
	<3a4j7onz.fsf@vps203.linuxvps.org>
	<718A5688-F55B-49D6-8371-243B1358AF46@clozure.com>
	<y6mbprrk.fsf@vps203.linuxvps.org>
	<hbsy7p7y.fsf@vps203.linuxvps.org>
	<9ab3e38b28bfdcd840af9091da075102.squirrel@mail.stardawn.org>
	<vdhdxb9m.fsf@vps203.linuxvps.org>
	<2192b07c4a8ef74499a6de2ec4b0a89e.squirrel@mail.stardawn.org>
	<skc71r31.fsf@vps203.linuxvps.org>
	<E8CD1C43-999C-4262-BA18-D9F8A1E0FC28@media.mit.edu>
	<k4xj1pus.fsf@vps203.linuxvps.org>
Message-ID: <4oomg0bk.fsf@vps203.linuxvps.org>

Quoth Sebastian Tennant <sebyte at smolny.plus.com>:
> Quoth Ian Eslick <eslick at media.mit.edu>:
>> What were the problems you encountered and solutions you found?   
>> Anything we should add to the FAQ?
>
> Nothing (reliable) that isn't already addressed in the FAQ.  I suspect there's
> something in git SBCL (1.0.32.xx) which was causing me problems on my 32-bit
> Debian box but I'll have to go back to that machine and check again before I
> can confirm this.

I can now confirm that bdb 4.7 stores in elephant 1.0 work fine with the stable
binary release of SBCL 1.0.32 in Debian.

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From reddaly at gmail.com  Sun Nov 22 13:04:03 2009
From: reddaly at gmail.com (Red Daly)
Date: Sun, 22 Nov 2009 05:04:03 -0800
Subject: [elephant-devel] More work on BDB interrupt-proofing
In-Reply-To: <c8226f6e985b2e0c6f6819f75981b10d.squirrel@mail.stardawn.org>
References: <c8226f6e985b2e0c6f6819f75981b10d.squirrel@mail.stardawn.org>
Message-ID: <f6700f0c0911220504p64f421f8w4e5c4664d6b5873f@mail.gmail.com>

Hi Leslie et al,

On Wed, Oct 14, 2009 at 4:37 AM, Leslie P. Polzer
<sky at viridian-project.de> wrote:
> The attached patch makes BDB more robust against interrupts in various
> places. It achieves this mainly by wrapping critical paths in
> WITHOUT-INTERRUPTS and by registering cursors on the Lisp level
> (and cleaning up any dangling ones when the txn ends).

Why do these sections require WITHOUT-INTERRUPTS instead of locking?
WITHOUT-INTERRUPTS is deprecated in Allegro, and in lisps running on
multiple cores, it seems to have no atomicity guarantees that are
immediately relevant to me.

Quoted below is the Allegro documentation.  I'm not sure if
WITHOUT-INTERRUPTS has the same semantics in other lisps:

"Because Symmetric Multiprocessing (SMP) will be introduced into
Allegro CL 9.0 on some platforms, use of multiprocessing macros
without-interrupts and sys:without-scheduling, whose behavior is
significantly different in an SMP environment, is deprecated,
particularly for those users who intend to use the SMP facility. See
smp.htm for full details, but in brief, without-interrupts and
sys:without-scheduling effectively provided object locking -- because
they prevented process switches, and only one Lisp process could ever
run at a time, no object could be modified by another process while
code wrapped in the macros was being run. In an SMP Lisp, that is no
longer true because processes can run on separate processors
simultaneously. Therefore, object locking must be achieved in some
other fashion. See smp.htm for links to object locking functionality."



>
> This is work in progress, although it passes the test suite.
>
> ?Leslie
>
> --
> http://www.linkedin.com/in/polzer
> _______________________________________________
> elephant-devel site list
> elephant-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/elephant-devel
>

Best,
Red Daly



From sky at viridian-project.de  Sun Nov 22 18:28:10 2009
From: sky at viridian-project.de (Leslie P. Polzer)
Date: Sun, 22 Nov 2009 19:28:10 +0100 (CET)
Subject: [elephant-devel] More work on BDB interrupt-proofing
In-Reply-To: <f6700f0c0911220504p64f421f8w4e5c4664d6b5873f@mail.gmail.com>
References: <c8226f6e985b2e0c6f6819f75981b10d.squirrel@mail.stardawn.org>
	<f6700f0c0911220504p64f421f8w4e5c4664d6b5873f@mail.gmail.com>
Message-ID: <01cb49adf67477b012f0450df8bff181.squirrel@mail.stardawn.org>


Red Daly wrote:

> Why do these sections require WITHOUT-INTERRUPTS instead of locking?
> WITHOUT-INTERRUPTS is deprecated in Allegro, and in lisps running on
> multiple cores, it seems to have no atomicity guarantees that are
> immediately relevant to me.
>
> Quoted below is the Allegro documentation.  I'm not sure if
> WITHOUT-INTERRUPTS has the same semantics in other lisps:

I don't know about Allegro but on SBCL for example locks and
without-interrupts are orthogonal. The latter is all about
asynchronous POSIX signals.

  Leslie

-- 
http://www.linkedin.com/in/polzer




From eslick at media.mit.edu  Mon Nov 23 19:02:16 2009
From: eslick at media.mit.edu (Ian Eslick)
Date: Mon, 23 Nov 2009 11:02:16 -0800
Subject: [elephant-devel] BDB problems and a proper FAQ
In-Reply-To: <k4xm49gl.fsf@vps203.linuxvps.org>
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<7212A9EA-9836-4714-81FE-B36AAD3FB83E@media.mit.edu>
	<k4xm49gl.fsf@vps203.linuxvps.org>
Message-ID: <14FC2998-2FCA-4DA5-9224-2630AC5EAD60@media.mit.edu>

I've posted this FAQ documentation online.  -Ian

On Nov 19, 2009, at 11:10 AM, Sebastian Tennant wrote:

> Hi Ian,
>
> Sorry for the late reply.  I have a rather annoying cold at the  
> moment.
>
> Quoth Ian Eslick <eslick at media.mit.edu>:
>> I'm ostensibly the primary developer on Elephant; the three prior  
>> primary
>> developers have been inactive for years now.  Leslie has recently  
>> been kind
>> enough to step up and help with user support and bug fixes and will  
>> certainly
>> be recognized in the 1.0 manual when we're ready for a final release
>> (updating the manual has been one of the gating tasks).  We all do  
>> our best
>> to help and want Elephant to be as robust as possible.  I in  
>> particular have
>> been lousy at responding recently, but there are of course good  
>> reasons for
>> that.
>
> I've no doubt there are.  Glad to hear that Leslie will be getting a  
> mention in
> the 1.0 manual.  He's been very patient with me!!!
>
>> Anyway, I'm sorry to hear that your experience has been so  
>> frustrating.
>> Unfortunately, the 64-bit lisp + OS + BDB combinations have made  
>> things much
>> more difficult than they once were.  As an aside, Franz  
>> AllegroCache has most
>> of the features of Elephant and has a great and responsive support  
>> staff.  If
>> you are frustrated with the state of free, volunteer-developed lisp  
>> libraries
>> it is possible to pay to get more reliability and better support!
>
> Noted.
>
>> There are dozens of users of Elephant, some casual and some  
>> serious.  Some
>> are working on older versions of the library as their deployed  
>> systems are
>> stable and not actively developed so not everyone is subject to  
>> problems with
>> the current release.
>
> Naturally.
>
>> Leslie and I are just the most vocal on this list.
>
> Understood.
>
>> I'm running Elephant under ClozureCL these days and it works fine  
>> under Snow
>> Leopard as well as Debian.
>
> Noted.
>
>> What's your current status, are you still stuck with the 'invalid  
>> argument'
>> problems?
>
> Yeah, I'm afraid so (on my Mac).  I've got an elephant-1.0 bdb 4.7  
> store
> working on a 32-bit Debian box with SBCL 1.0.30 but am having no  
> luck with the
> Mac, even with CCL.  I'll post a backtrace in a separate thread.
>
>> I think that a FAQ on the home page would be a great way to keep  
>> track of the
>> kind of issues that you have experienced.  Here is a first crack at  
>> it;
>> please add your own experiences/observations and I'll post it on  
>> the home
>> page.
>
> You've covered pretty much everything.  Please find attached HTML  
> (which I hope
> is a help).
>
> Regards,
>
> Sebastian
> -- 
> Emacs' AlsaPlayer - Music Without Jolts
> Lightweight, full-featured and mindful of your idyllic happiness.
> http://home.gna.org/eap
>
> <elephant-1.0.html>_______________________________________________
> elephant-devel site list
> elephant-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/elephant-devel




From sebyte at smolny.plus.com  Tue Nov 24 18:35:04 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Tue, 24 Nov 2009 18:35:04 +0000
Subject: [elephant-devel] BDB problems and a proper FAQ
References: <fx8jrb0h.fsf@vps203.linuxvps.org>
	<3a4jr6o8.fsf@vps203.linuxvps.org>
	<20091112183608.GA309@viridian-project.de>
	<ljia7rku.fsf@vps203.linuxvps.org>
	<7212A9EA-9836-4714-81FE-B36AAD3FB83E@media.mit.edu>
	<k4xm49gl.fsf@vps203.linuxvps.org>
	<14FC2998-2FCA-4DA5-9224-2630AC5EAD60@media.mit.edu>
Message-ID: <skc34vqf.fsf@vps203.linuxvps.org>

Quoth Ian Eslick <eslick at media.mit.edu>:
> I've posted this FAQ documentation online.

It looks good.  Thanks.

Seb
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap