[gps-devel] gps compile.

Arnaud Charlet charlet at ACT-Europe.FR
Tue Jun 29 15:40:11 CEST 2004


> For Debian, I had to copy several source files from GCC 3.4 into the
> GPS sources to get GPS to compile.  I did not have to recompile GNAT
> itself, but the effort required to get GPS to compile with GNAT 3.15p
> was not trivial at all (approximately 70 hours' work).  BTW, this

I suspect something is not quite right in the way you build things.

> means that the statement, on the GPS home page, that "you need GNAT >=
> 3.15" is not entirely true.

It is entirely true, I find your statement offending.

> I find it unfortunate that GPS will not compile out of the box with
> the current public release of GNAT, 3.15p.  I have had this problem

It does compile out of the box with GPS releases, I checked that personnally.
I find it unfortunate that people will more easily blame others rather than
themselves.

If it does not for you, it means that you are either:
- not following the instructions precisely enough
- building from CVS, which indeed is a little bit trickier, since GNAT
  sources are not packaged together with the CVS sources.
- or starting from a miscompiled GNAT (the instructions concerning GNAT 3.15p
  assume a proper binary, such as the ones available on the libre site).

As for the particular issue of the pretty printer package, that's indeed
an unfortunate issue, although it will soon become OBE for two
reasons:
- GCC 3.4 is out and handles/ignores this package as expected
- GPS 2.0 should be out soon and won't generate this package unless
  asked for explicitely

> Any one of these solutions would make life _much_ easier for everyone
> who wants to use this software and contribute to it, or for
> distribution maintainers like myself.

You will find than rather than demanding more work on our side, if you follow
more precisely the instructions, you will reduce the amount of work for
you, without having to demand anything. At least that's the case for GPS.

As for other packages, this mailing list is not the right place for discussing
them.

Arno


More information about the gps-devel mailing list