On Fri, 17 May 2013 13:40:13 +0200 Dominique Dumont <dod@debian.org> wrote: > This may be a dumb question, but looking at this example of different binNMU > changelogs [1], I wonder what is the point of having a log specific to each > build system ... Part of the problem is that binNMU's are not always done for all architectures supported by the package, only for problematic ones. So even if the changelog was precisely the same for all architectures rebuilt, Multi-Arch installs would still fail if one of the foreign architectures was *not* one of the binNMU targets and at least one other was. i.e. we need a solution which is not based entirely on making the changelog file itself the same when rebuilds are done. We can't afford to force every architecture to be rebuilt for every binNMU. (Not only because it's a waste of buildd resources, also because some binNMU's are done because timing issues meant that a particular arch broke in the test suite and doing a second rebuild could actually fail.) > In other word, can't we have a binNMU log generated by *the* system that will > trigger the rebuild ? So the log would be common for all binary packages. That's a dead end - binNMU's don't affect all architectures equally. -- Neil Williams ============= http://www.linux.codehelp.co.uk/
Attachment:
pgp9NHtXLyxB4.pgp
Description: PGP signature