This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: One version number, &c, take two
DJ Delorie <dj@redhat.com> writes:
>> > I was thinking about divergent opinions about the tagging process,
>> > not the actual date used. Binutils already has an alternate
>> > scheme with daily tagging.
>>
>> Could you expand a bit please?
>
> Binutils uses bfd/version.h for everyone's datestamp. Adding a second
> datestamp file would make their situation worse, not better.
After some shakeout time in gcc only, presumably they could start
using a similar scheme? (yes, I would help adapt scripts and
makefiles as necessary.)
>> > The toplevel config/* is currently manually sync'd between gcc and
>> > binutils, it would be a mess if that blew up my scripts.
>>
>> It's just the one file, could you special-case it?
>
> Not easily. The script works by inclusion; currently I include
> ./config. To exclude one file, I'd have to manually list (and
> maintain the list of) every other file in ./config/*.
>
> Could you commit to both repositories, so that the file stays in sync?
I'm not sure if I understand the requirement here. Would it be
necessary for the nightly GCC cron job to commit the file to each
repository on every bump? Or can I just create the file the first
time and let your sync scripts propagate the nightly bumps?
What about the other files, which change occasionally?
>> In my opinion, only files specific to the gcc program belong in the
>> gcc subdirectory.
>
> Yet, you're planning on putting something that's not globally used in
> a global directory. Too bad we don't have an in-between solution;
> something that's common to all the gcc components but separate from
> binutils and gdb. The gcc subdirectory used to be that.
I honestly don't see the problem here. There is lots of stuff in
toplevel, config, and include that is only relevant to a subset of the
projects in src+gcc.
zw
- References:
- One version number, &c, take two
- Re: One version number, &c, take two
- Re: One version number, &c, take two
- Re: One version number, &c, take two
- Re: One version number, &c, take two
- Re: One version number, &c, take two