Big fan of quilt here. The primary use cases for me are applying changes to different environments or running against different environments (local/stage/prod).
That means I keep several independent series of patch series, like: patches/series-flyway-local-oracle patches/series-flyway-local-postgre patches/series-flyway-stage patches/series-flyway-prod I don't set QUILT_SERIES as it only alters single environment (Bash). I what settings immediately available globally. So I wrote command "switch" that updates: $QUILT_PC/.quilt_series And enhanced bash_completion to complete series name: quilt switch series-flyway- TAB It is extremely convenient workflow, especially that Git doesn't have anything like Mercurial queues and it is the winner of the DVCS race. I added some guards, like you cannot "quilt switch" if some patches are applied. I'd like to have such feature in the core distribution because at least bash_completion script is overridden during package update and distributing my alteration to all my PCs is a bit a headache. What is your ideas about inclusion, naming the command (it can be "switch-series") and completion conventions (like absolute/relative path or anything prefixes "series" in the current QUILT_PATCHES)? In case if .pc is not initialized I think command should terminate with warning, recommending "quilt init". It is for interactive use so should be intuitive and simple to use. Unlike dealing with QUILT_SERIES (which is fine for scripting but do you have bash completion for undefined variables?). -- http://defun.work/ _______________________________________________ Quilt-dev mailing list [email protected] https://lists.nongnu.org/mailman/listinfo/quilt-dev
