[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: mod_caml and cocanwiki (was: Re: Cocanwiki package)



On Sat, Sep 11, 2004 at 03:13:54PM +0100, Richard Jones wrote:
> It's quite a complex tool so it comes in several parts.  The key ones
> are:
> 
>   mod_caml.so  - The module which is loaded into Apache, and contains
>                  all the C code along with apache.cmo, cgi.cmo,
>                  mod_caml.cmo.
> 
>   Cgi_escape, Template - CAML modules which are most useful inside
>                  mod_caml scripts, but could conceivably be used
>                  independently.
> 
>   Registry - An extension which is loaded into Apache (using the
>                  CamlLoad command) to support CGI scripts.

Thanks for the explanation. I therefore propose to have two packages
(probably this is exactly what you already did, but I'm offline and I
don't have an up-to-date copy of the SVN repo, see below, sorry).

The first one, named "libapache-mod-caml" following the apache mod
convention, will contain mod_caml.so, the registry and misc stuff like
the icons.

I'm a bit puzzled about the second one. I think it would be good to have
a "libtemplate-ocaml-dev" package since template processing is a useful
task even without mod_caml. But now I don't know what to do with
Cgi_escape, IMO it's not worth the effort to make a package just for it
since we already have ocamlnet which have similar features. I think we
could simply install Cgi_escape in /usr/lib/ocaml/3.08/apache/ and let
it be contained in the mod-caml package. In the future we can put there
other auxiliary modules of mod-caml as well.

What do you think of this naming/split?

Regarding the dependencies of the two package they should be independent
(no Depends), IMO you can make mod-caml package Suggests or Recommends
libtemplate-ocaml-dev depending on how much you consider it a useful
extension for mod-caml programmers. I would go for Recommends.

Regarding examples they should be put in the mod-caml package.

.cmi/.mli should be installed side by side with the package containing
corresponding ocaml objects.

> Actually, when I do this I get:
> $ lintian mod-caml_1.2.1-2.dsc 
> W: mod-caml source: cvsignore-file-in-source .cvsignore
<snip>
> These are definitely WON'T FIX :-) I like my .cvsignore files there.

Ok :), it's just a warning, but lintian is right: they really shouldn't
be there. debian or not debian, .cvsignore files should be in your
working copy but not in the tarball for the final user, if you think at
the autotool you usually have a "make dist" target which create a
tarball which does not contain CVS/ directories nor .cvsignore files.

> >   W: libapache-ocaml: image-file-in-usr-lib usr/lib/mod_caml/caml-icons/error.png
> >   W: libapache-ocaml: image-file-in-usr-lib usr/lib/mod_caml/caml-icons/ok.png
> Fixed.

BTW, where have you put them? (I can't look it by myself since the SVN
repository didn't work a few minutes ago while I was connected and thus
I was unable to update my local repository)

> Also I get:
> E: libapache-ocaml: unstripped-binary-or-object ./usr/lib/apache/1.3/mod_caml.so
> 
> Unfortunately I can't fix this.  Stripping mod_caml.so removes the
> CAML bytecode (there was discussion of this on caml-list a few weeks
> back), and that is obviously a showstopper.

Of course. You can ignore this, if you would like to avoid this error to
be reported you can also provide a lintian override, that is a file
shipped with your package which, once installed, tells to lintian that
he should ignore this error for your package. If you're interested,
lintian overrides are described in the lintian user manual
/usr/share/doc/lintian/lintian.html/index.html

> There are updates which fix everything I can see in SVN now.

I will look at them as soon as I will be able to update my SVN local
copy.

Cheers.

-- 
Stefano Zacchiroli -*- Computer Science PhD student @ Uny Bologna, Italy
zack@{cs.unibo.it,debian.org,bononia.it} -%- http://www.bononia.it/zack/
If there's any real truth it's that the entire multidimensional infinity
of the Universe is almost certainly being run by a bunch of maniacs. -!-



Reply to: