caml-list - the Caml user's mailing list
 help / color / mirror / Atom feed
From: Guillaume Rousse <Guillaume.Rousse@inria.fr>
Cc: caml-list@inria.fr
Subject: Re: [Caml-list] findlib help: how to specify dependencies against non-findlib packages ?
Date: Mon, 02 Oct 2006 17:39:03 +0200	[thread overview]
Message-ID: <45213297.9010205@inria.fr> (raw)
In-Reply-To: <1159800765.13405.34.camel@localhost.localdomain>

Gerd Stolpmann wrote:
> Am Montag, den 02.10.2006, 15:30 +0200 schrieb Guillaume Rousse:
>> Gerd Stolpmann wrote:
>>> Am Montag, den 02.10.2006, 11:24 +0200 schrieb Guillaume Rousse:
>>>> I'm trying to add findlibsupport for camlimages.
>>>>
>>>> It has a dependency on graphics, which is easily handled by adding a
>>>> require variable in META file, as graphics provides findlib support.
>>>>
>>>> However, camlimages can also get compiled with either lablgtk or
>>>> lablgtk2 support (not both of them, as they have conflicting symbols),
>>>> but none of those packages have findlib support. It means I can't
>>>> ressort to smart dependency computation. I may eventually use the
>>>> linkopts variable to add explicit linking flags (such as -I
>>>> place/to/lablgtk2 lablgtk.cma), but this seem to be only possible for
>>>> linking, not for compiling, whereas I also need to pass -I
>>>> place/to/lablgtk2 option during compilation.
>>>>
>>>> So, is this just not possible at all ?
>>> Of course, it is possible: ...
> 
>> I'm speaking of incoming camlimages 3.0.0 release, not of camlimages
>> packages in some distribution. I can't interfere with already existing
>> lablgtk/lablgtk2 installation, unless there is a way to do it privately
>> to camlimages itself.
> 
> Sorry, I misunderstood you.
> 
> In general, findlib assumes that prerequisite packages already have
> findlib support. There is no clean built-in way to cope with the
> situation, simply because if several packages depended on lablgtk and
> fixed the missing META file in their own way it would not be possible to
> resolve dependencies cleanly anymore.
> 
> What you can still do is to define a subpackage
> camlimages.lablgtk_substitute (syntax see below) and to depend on this
> subpackage if you see that lablgtk is installed without META file
> (otherwise just depend on lablgtk). This subpackage contains the
> substitute for the missing META file. However, as said this is no really
> clean solution but probably the best you can do.
The cleaner solution would be to push META support to lablgtk maintainer
directly. Does anyone already tried ?

> You can define subpackages in META files with this syntax:
> 
> <this is the META file for camlimages>
> package "lablgtk_substitute" (
>   <META directives>
> )
> 
> The subpackage can then be referred to as camlimages.lablgtk_substitute.
> For a complete example see the camlp4 META file coming with findlib. 
> 
> As lablgtk is usually installed below the standard library directory,
> you can refer to this directory using
> 
> directory = "+lablgtk"
> 
> Hope this helps.
That was exactly what I needed, thanks. I just have to handle the
conditional META support in lablgtk now.

> You might also be interested in how GODI generates a
> META file for the current version of camlimages (it assumes, however,
> that lablgtk is installed _with_ META file):
> 
> https://gps.dynxs.de/svn/godi-build/trunk/godi/godi-camlimages/create-META
Well, I hope this won't be necessary anymore with next release, if I
understood correctly how to generate it myself.

You're welcome to checkout current CVS version of camlimages to check.
-- 
Guillaume Rousse
Projet Estime, INRIA
Domaine de Voluceau
Rocquencourt - B.P. 105
78153 Le Chesnay Cedex - France


      reply	other threads:[~2006-10-02 15:39 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-10-02  9:24 Guillaume Rousse
2006-10-02 10:33 ` [Caml-list] " Gerd Stolpmann
2006-10-02 13:30   ` Guillaume Rousse
2006-10-02 14:52     ` Gerd Stolpmann
2006-10-02 15:39       ` Guillaume Rousse [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=45213297.9010205@inria.fr \
    --to=guillaume.rousse@inria.fr \
    --cc=caml-list@inria.fr \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).