Re: Differences between gnat 4.4.5 and 4.5.1

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

Mart van de Wege writes:
> Julian Leyh <julian@vgai.de> writes:
>
>> On 10 Mrz., 07:01, Noel Duffy <hobcraft-use...@yahooxtra.co.nz> wrote:
>>> I wonder whether what I am seeing is a result of a difference in gcc-gnat
>>> version, a difference in distribution packaging, or possibly a
>>> combination of both. Is there anyone using Fedora here who could possibly
>>> shed any light on this?
>>
>> What do "file" and "ldd" tell you about them? Are the sizes more
>> similar if you "strip" them?
>
> I'd be inclined to think stripping may be the culprit. Debian is well
> known to use stripped binaries by default, so I wouldn't be surprised
> if the various toolchains are set up that way as well.
>
> IIRC, this is what led them to stop being the default GNU distro, as
> the GNU folks like their binaries with debug info by default. And
> since Fedora is mostly a Red Hat project, and Red Hat has absorbed a
> *lot* of the GCC devs via Cygnus, I'd expect Fedora to *not* strip by
> default.
>
> Ludovic could probably answer that one in his sleep.

No, not in my sleep :)

Stripping means removing debugging symbols from an executable.  For the
debugging symbols to be present in the first place requires an explicit
-g option.

The "default" you speak of is that contained in most upstream Makefiles.
Debian *packaging scripts*, by default, passes -g to the compiler and
then strips the executables, moving the debugging symbols to a separate
-dbg package (this is in fact mandatory for the Ada libraries...).  But
packaging scripts have nothing to do with the OP's way of calling GCC.

Personally I'd be inclined to think the difference in size is due to the
compiler, not differences between distributions.  I'd have to check with
gnat-4.5 when it reaches a usable state.

-- 
Ludovic Brenta.