Re: ada-mode and GNAT project files (was: GVD)

[Ludovic Brenta <ludovic@ludovic-brenta.org>]

Emmanuel Briot writes:
> Hi there,
>
> A quick update on the status of ada-mode:
> AdaCore is no longer actively working on developing ada-mode, although
> we are still fixing bugs in it (in particular to keep it up-to-date with
> regards to the cross-references generated by GNAT in the ALI files).
>
> I'll try and take the time to contribute the latest version to FSF for
> inclusing in Emacs (I am still the official ada-mode maintainer for
> Emacs, although I would gladly give over that job if someone else is
> motivated -- I no longer use Emacs myself, and feel like I might be
> restraining the development of ada-mode).
>
> As far as I know, the FSF version should contain almost the latest
> version. In any case, the version released with later versions of GNAT
> (or Glide, as our package is called) are the latest, and what I will
> contribute to the FSF, so I would encourage you to start from there).

Well, emacs 21.4 comes with an ada-mode which is quite different from
the ada-mode 3.6 on libre.adacore.com.  Quick illustration from my system:

$ diff -u \
  /usr/share/emacs/21.4/lisp/progmodes/ada-mode.el \
  /usr/share/emacs/site-lisp/ada-mode/ada-mode.el | wc -l
1979

That's 1979 lines of unified diff :(

Another problem is that emacs releases are very few and far between.

> By any means keep using the ada-mode name in Emacs, no need to fork
> here. This is a win-win situation potentially (AdaCore will
> contribute now and then some patches, and benefit from the patches
> you might make).

This seems to be the best way forward, except for people who would
like to use my patch sooner rather than later :)

> With regards to supporting the .gpr files: Glide contains a small
> Ada binary that parses the GPR file, and outputs a temporary .adp
> file that Emacs reads on the fly to get the list of source dirs and
> object dirs.  This is certainly the most efficient way to do, since
> parsing the .gpr files is a complex task, that I doubt your parser
> fully does (there are lots of features in there). It is better to
> reuse the same parser that GNAT itself uses.

There is a lot of merit to this.  Where is this Ada executable?  Or
rather, where are the sources for it?

My patch is quite simplistic, as it only cares about Source_Dirs and
Object_Dir.  But it does perform variable substitution there :)

Also I have been toying with the idea of creating a shared library
containing the project-parsing code from GNAT.  Several tools could
then make use of that shared library; GPS is one of them.

-- 
Ludovic Brenta.