Re: empty TIMESTAMP, stamp-vcl

[Gary V. Vaughan]

Hi Eric,

On 28 Mar 2007, at 12:41, Eric Blake wrote:
> I still hope to incorporate the ideas in libtool so that m4 no longer
> requires GNU make
> (http://lists.gnu.org/archive/html/libtool-patches/2007-03/ 
> msg00048.html).

Done :-D

> In the meantime, I have been playing with git, which does not  
> expand RCS
> keywords, and that exposed a testsuite bug.  So part of fixing  
> stamp-vcl
> will be figuring out how to have a sane TIMESTAMP variable  
> regardless of
> whether the repository is from CVS or from git (I'm thinking of having
> mkstamp recognize whether m4/CVS or m4/.git exists, and continue to  
> use
> the RCS Revision of ChangeLog in CVS, but the first few characters  
> of the
> hash in 'git ls-tree HEAD ChangeLog' in git).

Consider that the timestamp is there to allow us to pull the exact  
version
a bug-reporter used by reference to the version string we hope they  
quote.
It doesn't actually matter what we use for that string, just that we can
get back to the same build that generated the bug-report.

There are two implications:
  i) if we master from git, then the git revision hash gives us that,
     as long as we push that information into the CVS mirror.
 ii) except that doesn't tell us which gnulib release was used, so we
     need to factor the gnulib timestamp information too :-(  I think
     we can maybe get away with just making bootstrap write a gnulib
     repo timestamp into a version.c file...

On the other hand maybe we can persuade the gnulib folks to have a  
proper
release schedule, or else add something to gnulib-tool to help us keep
track of the particular non-releases the bug-reporter bootstrapped with?
I'm afraid that my desperate idea of forking gnulib at m4 feature freeze
time, cherry picking gnulib bug fixes to the modules we care about into
the fork and making a gnulib-for-m4-2.0 pseudo-release would make things
even harder here.

Cheers,
        Gary
--
  ())_.              Email me: address@hidden
  ( '/           Read my blog: http://blog.azazil.net
  / )=         ...and my book: http://sources.redhat.com/autobook
`(_~)_ Join my AGLOCO Network: http://www.agloco.com/r/BBBS7912

PGP.sig
Description: This is a digitally signed message part