Hi Markus, [Markus Oberhumer] > > > Please review and apply if you find it useful. > > > (...) > > > [PATCH] Only touch the .timestamp file if needed.
[Jean Delvare] > > I fail to see the benefit, can you please explain? [Markus Oberhumer] > well, it's the the correct thing to do. I'm not convinced, sorry ;) The timestamp means "at that time, the patch was up-to-date". In that respect the current implementation is correct. And so is yours, but why change something that works? > But the main reason is that I often use "quilt refresh" as a substitute for > a missing "quilt status", and this way it will better work on read-only > mounts - at least if the patch is unchanged. Using quilt on read-only mounts doesn't make much sense to me. Or at least you wouldn't run "quilt refresh" in these conditions. And your fix hardly addresses this problem, as if a patch needs to be refreshed you'll still have an error. If you need "quilt status" maybe it would be better to implement this. I remember this was discussed once already [1], and the problem was that not everyone agreed on what information quilt status should print. Maybe you want instead a flag to quilt refresh which would emulate the refresh but not actually do it. E.g.: $ quilt refresh --needed Patch $patch would need to be refreshed $ Just an idea though. [1] http://lists.gnu.org/archive/html/quilt-dev/2006-01/msg00020.html -- Jean Delvare _______________________________________________ Quilt-dev mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/quilt-dev
