This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Patch to gcc_update to include revision number in LAST_UPDATED


Mark Mitchell <mark@codesourcery.com> wrote:

>>> 2) In bug reports, if gcc -v says "105678M" (note the 'M'), we
>>> would know there are local patches in the tree, which is a typical
>>> msitake done even by experienced GCC maintainers. "Hey bootstrap is
>>> broken", "not for me", "not for me", "not here", "ah sorry, I had
>>> local patches".
>
> If we start to do this (i.e., generate version information
> dynamically),
> it goes without saying that it should still do something sensible if
> you're not using subversion -- like if you're building a release.
>
> Also, I've been planning for a while to add an option to configure to
> set the part of the version information other than the raw version
> number, so that you could do something like
>
>   configure --distribution="CodeSourcery Build #27 [20051125]"
>
> and get binaries that identified themselves as such.  Right now, you
> have to modify version.c, and that's a hassle -- and, then, if people
> rebuild your sources with modifications, the modified sources still
> claim to be your releases, etc.
>
> I'm not sure how this interacts with the suggestion above, but I think
> we should make sure that neither things prevents the other.


I think they're pretty orthogonal: my proposal would affect the raw version
number, while yours the identifier tag. I don't have enough shell-fu to
implement either, sadly.

Giovanni Bajo


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]