From gnurag at gmail.com  Tue Sep 18 14:35:21 2007
From: gnurag at gmail.com (Anurag)
Date: Tue, 18 Sep 2007 20:05:21 +0530
Subject: [fetter-devel] generating bindings for librrd
Message-ID: <5a383a5b0709180735u78515c95yfb25085a4012f5ac@mail.gmail.com>

Hello,

I've been trying to generate bindings for librrd using verrazano. I
added the header file entries for librrd in generate-bindings and ran
the file to get the librrd-bindings file.

However when i try to include it in a package and load the package, i
get this error mesage.

---
Unknown CFFI type: LIBRRD-CFFI-BINDINGS::_-OFF-T.
[Condition of type SIMPLE-ERROR]
----

Any idea where i am going wrong? I am using AllegroCL 8.0 on an amd64
box. I'll be happy to provide more information if required.

Anurag

-- 
-- 
Anurag <0xB20A82C1>


From attila.lendvai at gmail.com  Tue Sep 18 15:12:07 2007
From: attila.lendvai at gmail.com (Attila Lendvai)
Date: Tue, 18 Sep 2007 17:12:07 +0200
Subject: [fetter-devel] generating bindings for librrd
In-Reply-To: <5a383a5b0709180735u78515c95yfb25085a4012f5ac@mail.gmail.com>
References: <5a383a5b0709180735u78515c95yfb25085a4012f5ac@mail.gmail.com>
Message-ID: <b54b34af0709180812w1052dd45gf9c207a19361914f@mail.gmail.com>

> Any idea where i am going wrong? I am using AllegroCL 8.0 on an amd64

darcs pull cffi

-- 
 attila


From christian.lynbech at tietoenator.com  Wed Sep 19 14:39:59 2007
From: christian.lynbech at tietoenator.com (Lynbech Christian)
Date: Wed, 19 Sep 2007 16:39:59 +0200
Subject: [fetter-devel] generating bindings for librrd
In-Reply-To: <5a383a5b0709180735u78515c95yfb25085a4012f5ac@mail.gmail.com>
	(Anurag's message of "Tue, 18 Sep 2007 20:05:21 +0530")
References: <5a383a5b0709180735u78515c95yfb25085a4012f5ac@mail.gmail.com>
Message-ID: <ytqpir664t34.fsf@ul000131.eu.tieto.com>

>>>>> "Anurag" == Anurag  <gnurag at gmail.com> writes:

Anurag> ---
Anurag> Unknown CFFI type: LIBRRD-CFFI-BINDINGS::_-OFF-T.
Anurag> [Condition of type SIMPLE-ERROR]
Anurag> ----

This looks like a problem with derived types that I have been battling
with the last time I tried to use Verranzano (which is like a year back
so things may have changed).

Suppose you have a type in the manner of (C syntax not precise):

        struct {
          ...
          enum YYY { ...}
        } XXX

you are effectively defining a new type YYY within the definition of
XXX. Verranzano does handle this but to avoid nameclashes, it adds a
running number to these definitions such that you get something like

        (defctype YYY12345 ...)

but usages of the type will stil refer to YYY which is defined nowhere.

Try searching the generated CFFI bindings for OFF-T to see whether this
is indeed the problem. If it the case, I believe that you can easily
define a new CFFI type as an alias pointing _-OFF-T to _OFF-Txyz as
necessary, but beware that this must be updated every the number
changes. The version I was using did not reset the number variable
between runs so it would change each and every time you called
generate-bindings. This is easily fixable but the numbwer will still
change every time anything changes within the collection of header files
involved in the generation.

A better solution would be to change the name generation such that the
first such an internal type is met, it would be named "YYY" and if
another internal type was encountered using the same name, that would be
named "YYY2". I think that would catch the vast majority of
cases. Alternatively one should fix all usages of the type YYY to refer
to the correct calculated name which I think would be very hard to do.

As I said, things might have changed in current versions of Verranzano.


------------------------+-----------------------------------------------------
Christian Lynbech       | christian #\@ defun #\. dk
------------------------+-----------------------------------------------------
Hit the philistines three times over the head with the Elisp reference manual.
                                        - petonic at hal.com (Michael A. Petonic)




From gnurag at gmail.com  Thu Sep 20 10:53:34 2007
From: gnurag at gmail.com (Anurag)
Date: Thu, 20 Sep 2007 16:23:34 +0530
Subject: [fetter-devel] generating bindings for librrd
In-Reply-To: <ytqpir664t34.fsf@ul000131.eu.tieto.com>
References: <5a383a5b0709180735u78515c95yfb25085a4012f5ac@mail.gmail.com>
	<ytqpir664t34.fsf@ul000131.eu.tieto.com>
Message-ID: <5a383a5b0709200353u19811d59rbaa8d11391e27fe9@mail.gmail.com>

On 9/19/07, Lynbech Christian <christian.lynbech at tietoenator.com> wrote:
> This looks like a problem with derived types that I have been battling
> with the last time I tried to use Verranzano (which is like a year back
> so things may have changed).

Hi!

You are right, things have changed a lot in the cffi's darcs
repository. As per Attila's suggestion i used the latest version of
CFFI from the darcs tree, and the issue with OFF-T seems to be solved.

Now, when i am calling any foreign function which does not require any
parameter, it returns me proper string output (for example calling a
function which returns version of library)

But the foreign function doesnt seem to do work when i use normal lisp strings.

(rrd-last "/path/to/test.rrd" 1)

gives me,

The value "/path/to/test.rrd"  is not of type
(SIGNED-BYTE 32)


I understand that i should use,
(cffi:with-foreign-string (rrdfile "/path/to/test.rrd")
    (rrd-last rrdfile 1))

i still get,
The value #.(SB-SYS:INT-SAP #X08185508) is not of type (SIGNED-BYTE 32).
   [Condition of type TYPE-ERROR]


Is there something i can do to convert the string to 32byte signed type?

Anurag

-- 
Anurag <0xB20A82C1>


From attila.lendvai at gmail.com  Thu Sep 20 13:14:58 2007
From: attila.lendvai at gmail.com (Attila Lendvai)
Date: Thu, 20 Sep 2007 15:14:58 +0200
Subject: [fetter-devel] generating bindings for librrd
In-Reply-To: <5a383a5b0709200353u19811d59rbaa8d11391e27fe9@mail.gmail.com>
References: <5a383a5b0709180735u78515c95yfb25085a4012f5ac@mail.gmail.com>
	<ytqpir664t34.fsf@ul000131.eu.tieto.com>
	<5a383a5b0709200353u19811d59rbaa8d11391e27fe9@mail.gmail.com>
Message-ID: <b54b34af0709200614n6aa77e1yac7514ffe1c16303@mail.gmail.com>

> I understand that i should use,
> (cffi:with-foreign-string (rrdfile "/path/to/test.rrd")
>     (rrd-last rrdfile 1))
>
> i still get,
> The value #.(SB-SYS:INT-SAP #X08185508) is not of type (SIGNED-BYTE 32).
>    [Condition of type TYPE-ERROR]

last time i checked it worked for me with cffi and babel from
http://common-lisp.net/~loliveira/soc07/

-- 
 attila