From randall at randallsquared.com  Thu Sep 16 23:26:45 2004
From: randall at randallsquared.com (Randall Randall)
Date: Thu, 16 Sep 2004 19:26:45 -0400
Subject: [cl-prevalence-devel] Using strings as root keys?
Message-ID: <DFE37C21-0837-11D9-807D-000A95A0F1E8@randallsquared.com>

Hi, all.

I'm just starting to play with cl-prevalence.

Is there any reason (besides performance, which
I know would suffer with EQUAL or EQUALP) why
the root-objects hash in a system is EQ?


--
Randall Randall <randall at randallsquared.com>
'I say we put up a huge sign next to the Sun that says
"You must be at least this big (insert huge red line)
  to ride this ride".' -- tghdrdeath at hotmail.com




From scaekenberghe at common-lisp.net  Fri Sep 17 18:35:26 2004
From: scaekenberghe at common-lisp.net (Sven Van Caekenberghe)
Date: Fri, 17 Sep 2004 20:35:26 +0200
Subject: [cl-prevalence-devel] Using strings as root keys?
In-Reply-To: <DFE37C21-0837-11D9-807D-000A95A0F1E8@randallsquared.com>
References: <DFE37C21-0837-11D9-807D-000A95A0F1E8@randallsquared.com>
Message-ID: <57C7B936-08D8-11D9-ADD3-000A95B3E0D2@common-lisp.net>

On 17 Sep 2004, at 01:26, Randall Randall wrote:

> Hi, all.
>
> I'm just starting to play with cl-prevalence.
>
> Is there any reason (besides performance, which
> I know would suffer with EQUAL or EQUALP) why
> the root-objects hash in a system is EQ?

I honestly can't remember why I did this, but in CL using symbols as 
keys is free (as in doesn't cost you anything) *and* it is efficient.
I generally use 'managed prevalence' where the root keys are class 
names (and thus symbols).
Why would you want to use a string instead of a symbol ?

Sven




From randall at randallsquared.com  Fri Sep 17 19:33:16 2004
From: randall at randallsquared.com (Randall Randall)
Date: Fri, 17 Sep 2004 15:33:16 -0400
Subject: [cl-prevalence-devel] Using strings as root keys?
In-Reply-To: <9BD33738-08D7-11D9-ADD3-000A95B3E0D2@beta9.be>
References: <DFE37C21-0837-11D9-807D-000A95A0F1E8@randallsquared.com>
	<9BD33738-08D7-11D9-ADD3-000A95B3E0D2@beta9.be>
Message-ID: <6C23773C-08E0-11D9-AAEC-000A95A0F1E8@randallsquared.com>

On Sep 17, 2004, at 2:30 PM, Sven Van Caekenberghe wrote:

> Why would you want to use a string instead of a symbol ?

Possibly because I didn't understand that the root objects
are supposed to be containers for most of the actual data
structures, rather than those data structures directly.

I actually don't see any need for that now.  I should have
read more source before asking questions. :)

Thanks!

--
Randall Randall <randall at randallsquared.com>
Property law should use #'EQ , not #'EQUAL .




From randall at randallsquared.com  Sun Sep 19 09:09:21 2004
From: randall at randallsquared.com (Randall Randall)
Date: Sun, 19 Sep 2004 05:09:21 -0400
Subject: [cl-prevalence-devel] Another simple question
Message-ID: <97D224E0-0A1B-11D9-B94B-000A95A0F1E8@randallsquared.com>


In looking through managed-prevalence.lisp, it
appears that the "unique" id counter isn't saved
in the transaction log.  Is this an oversight,
or a feature I don't understand? :)

The relevant section is copied in below:

(defun tx-create-id-counter (system)
   "Initialize the id counter to 0"
   (setf (get-root-object system :id-counter) 0))

(defmethod next-id ((system prevalence-system))
   "Increment and return the next id"
   (incf (get-root-object system :id-counter)))

--
Randall Randall <randall at randallsquared.com>
Property law should use #'EQ , not #'EQUAL .




From randall at randallsquared.com  Sun Sep 19 22:15:59 2004
From: randall at randallsquared.com (Randall Randall)
Date: Sun, 19 Sep 2004 18:15:59 -0400
Subject: [cl-prevalence-devel] Fwd: Another simple question
Message-ID: <7C74A104-0A89-11D9-BDC4-000A95A0F1E8@randallsquared.com>

I wrote:
> In looking through managed-prevalence.lisp, it
> appears that the "unique" id counter isn't saved
> in the transaction log.  Is this an oversight,
> or a feature I don't understand? :)

The latter, it would appear, since next-id is only
called from inside a transaction.  My apologies
for the noise.

--
Randall Randall <randall at randallsquared.com>
Property law should use #'EQ , not #'EQUAL .




From scaekenberghe at common-lisp.net  Mon Sep 20 06:50:02 2004
From: scaekenberghe at common-lisp.net (Sven Van Caekenberghe)
Date: Mon, 20 Sep 2004 08:50:02 +0200
Subject: [cl-prevalence-devel] Fwd: Another simple question
In-Reply-To: <7C74A104-0A89-11D9-BDC4-000A95A0F1E8@randallsquared.com>
References: <7C74A104-0A89-11D9-BDC4-000A95A0F1E8@randallsquared.com>
Message-ID: <4BE95C87-0AD1-11D9-BD46-000A95B3E0D2@common-lisp.net>

On 20 Sep 2004, at 00:15, Randall Randall wrote:

> I wrote:
>> In looking through managed-prevalence.lisp, it
>> appears that the "unique" id counter isn't saved
>> in the transaction log.  Is this an oversight,
>> or a feature I don't understand? :)
>
> The latter, it would appear, since next-id is only
> called from inside a transaction.  My apologies
> for the noise.
>

Indeed.

It is however necessary to call (execute-transaction 
(tx-create-id-counter)) once before starting to use a newly created 
prevalence system for the first time, otherwise the counter is 
uninitialized.

Sven




From randall at randallsquared.com  Mon Sep 27 06:47:28 2004
From: randall at randallsquared.com (Randall Randall)
Date: Mon, 27 Sep 2004 02:47:28 -0400
Subject: [cl-prevalence-devel] problems with restoration (long)
Message-ID: <18ED236A-1051-11D9-B422-000A95A0F1E8@randallsquared.com>

So, I've been using cl-prevalence, and like it,
mostly. :)  I have 3 systems active in my application
(with at least 2 more expected before it will be
completely converted).

Two of them restore correctly.  One reliably does
not, and while the fact that two work okay suggests
that the problem is in my code, I can't find one.

When I change prevailed objects, they seem to update
in the log.  However, deeply nested objects seem not
to be restored correctly.

The classes involved are:

(defclass page (object-with-id)
   ((name :initarg :name :initform "" :accessor name)
    (title :initarg :title :initform () :accessor title))
   (:documentation ""))

(defclass xhtml-page (page)
   ((xhtml :initarg :xhtml :initform () :accessor xhtml)))

(defclass classroom (object-with-id)
   ((cover :initarg :cover :initform "" :accessor cover)
    (toc   :initarg :toc :initform () :accessor toc)
    (pages :initarg :pages :initform (make-hash-table :test 'equal)  
:accessor pages))
   (:documentation ""))

(defclass course (accessible)
   ((name          :initarg :name :initform "" :accessor name)   ; used  
as a human-readable id
    (title         :initarg :title :initform "" :accessor title) ; shown  
to users as the title of the course
    (classroom     :initarg :classroom :initform () :accessor classroom)
    (menu          :initarg :menu :initform () :accessor menu))
   (:documentation ""))

The "accessible" class just has some permissions control, which
works okay; I'm sure it's not essential to this problem.

When I create a course, giving it a classroom, with a single
xhtml-page in the pages hash, this is the XML output:

<OBJECT ID="1" CLASS="CL-PREVALENCE::TRANSACTION"><SLOT  
NAME="CL-PREVALENCE::ARGS"><NULL/></SLOT><SLOT  
NAME="CL:FUNCTION"><SYMBOL>CL-PREVALENCE::TX-CREATE-ID-COUNTER</ 
SYMBOL></SLOT></OBJECT>
<OBJECT ID="1" CLASS="CL-PREVALENCE::TRANSACTION"><SLOT  
NAME="CL-PREVALENCE::ARGS"><SEQUENCE ID="2" CLASS="CL:LIST"  
SIZE="3"><SYMBOL>COM.RANDALLSQUARED.COOL::COURSE</SYMBOL><SYMBOL>CL- 
PREVALENCE::NAME</SYMBOL><SYMBOL>CL:EQUALP</SYMBOL></SEQUENCE></ 
SLOT><SLOT  
NAME="CL:FUNCTION"><SYMBOL>CL-PREVALENCE::TX-CREATE-OBJECTS-S
LOT-INDEX</SYMBOL></SLOT></OBJECT>
<OBJECT ID="1" CLASS="CL-PREVALENCE::TRANSACTION"><SLOT  
NAME="CL-PREVALENCE::ARGS"><SEQUENCE ID="2" CLASS="CL:LIST"  
SIZE="2"><SYMBOL>COM.RANDALLSQUARED.COOL::XHTML-PAGE</SYMBOL><SEQUENCE  
ID="3" CLASS="CL:LIST" SIZE="2"><SEQUENCE ID="4" CLASS="CL:LIST"  
SIZE="2"><SYMBOL>COM.RANDALLSQUARED.COOL::TITLE</SYMBOL><SEQUENCE ID
="5" CLASS="CL:LIST"  
SIZE="1"><STRING>default-page-title</STRING></SEQUENCE></ 
SEQUENCE><SEQUENCE ID="6" CLASS="CL:LIST"  
SIZE="2"><SYMBOL>COM.RANDALLSQUARED.COOL::XHTML</SYMBOL><SEQUENCE  
ID="7" CLASS="CL:LIST"  
SIZE="1"><STRING>default-page-content</STRING></SEQUENCE></SEQUENCE></ 
SEQUENCE></SEQUENCE></SLOT><SLOT NAME="CL
:FUNCTION"><SYMBOL>CL-PREVALENCE::TX-CREATE-OBJECT</SYMBOL></SLOT></ 
OBJECT>
<OBJECT ID="1" CLASS="CL-PREVALENCE::TRANSACTION"><SLOT  
NAME="CL-PREVALENCE::ARGS"><SEQUENCE ID="2" CLASS="CL:LIST"  
SIZE="1"><SYMBOL>COM.RANDALLSQUARED.COOL::CLASSROOM</SYMBOL></ 
SEQUENCE></SLOT><SLOT  
NAME="CL:FUNCTION"><SYMBOL>CL-PREVALENCE::TX-CREATE-OBJECT</SYMBOL></ 
SLOT></OBJECT>
<OBJECT ID="1" CLASS="CL-PREVALENCE::TRANSACTION"><SLOT  
NAME="CL-PREVALENCE::ARGS"><SEQUENCE ID="2" CLASS="CL:LIST"  
SIZE="2"><SYMBOL>COM.RANDALLSQUARED.COOL::COURSE</SYMBOL><SEQUENCE  
ID="3" CLASS="CL:LIST" SIZE="3"><SEQUENCE ID="4" CLASS="CL:LIST"  
SIZE="2"><SYMBOL>COM.RANDALLSQUARED.COOL::CLASSROOM</SYMBOL><OBJECT  
ID="5" CLASS="COM.RANDALLSQUARED.COOL::CLASSROOM"><SLOT  
NAME="CL-PREVALENCE::ID"><INT>2</INT></SLOT><SLOT  
NAME="COM.RANDALLSQUARED.COOL::COVER"><STRING></STRING></SLOT><SLOT  
NAME="COM.RANDALLSQUARED.COOL::TOC"><NULL/></SLOT><SLOT  
NAME="COM.RANDALLSQUARED.COOL::PAGES"><HASH-TABLE ID="6"  
TEST="CL:EQUAL"  
SIZE="65"><ENTRY><KEY><STRING></STRING></KEY><VALUE><SEQUENCE ID="7"  
CLASS="CL:LIST" SIZE="1"><OBJECT ID="8"  
CLASS="COM.RANDALLSQUARED.COOL::XHTML-PAGE"><SLOT  
NAME="CL-PREVALENCE::ID"><INT>1</INT></SLOT><SLOT  
NAME="CL-PREVALENCE::NAME"><STRING></STRING></SLOT><SLOT  
NAME="COM.RANDALLSQUARED.COOL::TITLE"><SEQUENCE ID="9" CLASS="CL:LIST"  
SIZE="1"><STRING>default-page-title</STRING></SEQUENCE></SLOT><SLOT  
NAME="COM.RANDALLSQUARED.COOL::XHTML"><SEQUENCE ID="10" CLASS="CL:LIST"  
SIZE="1"><STRING>default-page-content</STRING></SEQUENCE></SLOT></ 
OBJECT></SEQUENCE></VALUE></ENTRY></HASH-TABLE></SLOT></OBJECT></ 
SEQUENCE><SEQUENCE ID="11" CLASS="CL:LIST"  
SIZE="2"><SYMBOL>COM.RANDALLSQUARED.PSIUSER::HANDLES</SYMBOL><SEQUENCE  
ID="12" CLASS="CL:LIST"  
SIZE="2"><INT>78</INT><INT>75</INT></SEQUENCE></SEQUENCE><SEQUENCE  
ID="13" CLASS="CL:LIST"  
SIZE="2"><SYMBOL>CL-PREVALENCE::NAME</SYMBOL><STRING>course</STRING></ 
SEQUENCE></SEQUENCE></SEQUENCE></SLOT><SLOT  
NAME="CL:FUNCTION"><SYMBOL>CL-
PREVALENCE::TX-CREATE-OBJECT</SYMBOL></SLOT></OBJECT>

So far, so good, I think.  In case it's not clear from the
above (it isn't to me), the course has id 3, the classroom id
2, and the xhtml-page id 1, since the default contents get
created just before they're passed to tx-create-object in
managed-prevalence.lisp.

But upon restarting the system, while all the id integers are
correct, the xhtml-page isn't attached to the pages hash
anymore:
* (find-courses) ; a standin for (find-all-objects *system* 'course)

(#<COURSE {49115E15}>)
T
* (get-id (car (find-courses)))

3
* (get-id (classroom (car (find-courses))))

2
* (pages (classroom (car (find-courses))))

#<EQUAL hash table, 1 entry {491146FD}>
* (gethash "" (pages (classroom (car (find-courses)))))

NIL
NIL
* (find-object-with-id *system* 'xhtml-page 1)

#<XHTML-PAGE {49110995}>
T

So, the obvious question is, how come the hash doesn't
get it's xhtml-page back?



--
Randall Randall <randall at randallsquared.com>
"And no practical definition of freedom would be complete
  without the freedom to take the consequences. Indeed, it
  is the freedom upon which all the others are based."
  - Terry Pratchett, _Going Postal_