On Tue, Feb 03, 2015 at 04:38:15PM +0000, Ian Campbell wrote: > On Tue, 2015-02-03 at 10:58 -0500, Kevin O'Connor wrote: > > On Tue, Feb 03, 2015 at 01:23:53PM +0100, Olaf Hering wrote: > > > The timestamp and hostname carries no value for the default build. If > > > its really required in special environment, or if someone is emotionally > > > attached to such strings, it is is handled by existing logic a few lines > > > above: provide your own .version file > > > > > > Once this patch is applied it is possible to get reproducible builds of > > > Xens hvmloader. > > > > I've found the version information to be quite helpful when we get > > trouble reports - it helps indicate what was being run and who built > > the code (eg, distribution or local). During development cycles, it > > helps verify that test runs are executed with the expected code. > > > > What breaks if the version changes? > > > > It should be possible to force a static version with "make > > VERSION='xyz'", but I'd ask that that not be done for production > > builds as it makes handling trouble reports harder. > > Olaf posted some similar patches against Xen and it was suggested that > using the git HEAD time and/or "git describe" would help. > > We could probably arrange to do this when building SeaBIOS as part of > Xen, or for tarball rather than git builds use the Xen release version?
SeaBIOS uses "git describe" as part of its version string today, and for SeaBIOS tar releases we embed a ".version" file with the release info (see scripts/buildversion.sh and scripts/tarball.sh for details). Even with this, the build timestamp is still useful during development cycles (when the head may be dirty). And the hostname/timestamp are useful as a way of backtracking to the gcc/binutils environment on trouble reports even from official releases. -Kevin _______________________________________________ SeaBIOS mailing list [email protected] http://www.seabios.org/mailman/listinfo/seabios
