From: Felipe Micaroni Lalli Date: Sat, 03 Oct 2015 06:44:31 GMT Subject: Re: Git feature request: mark a commit as minor Message-ID: <560F794F.2010809@walltime.info> In-Reply-To: Thank you Jacob. Actually we already use the keyword MINOR for that, exactly as you said. The suggestion was made because I think it is a common behavior and it would be nice to be a meta info to standardize this (today each team adopt a different pattern for that - you used "TRIVIAL" e.g.). Nice things could be done with this meta-info. It could be totally ignored (current git operation) or it could be used to filter, to sort, to group commits, to show the log pretty etc. > The issue is that not everyone considers these changes as "minor". I understand this issue, I know it is subjective. But if someone don't want to make the distinction just don't use the argument --hide-minor for example. On 03/10/2015 03:17, Mikael Magnusson wrote: > On Sat, Oct 3, 2015 at 8:11 AM, Jacob Keller wrote: >> On Fri, Oct 2, 2015 at 2:38 PM, Felipe Micaroni Lalli >> wrote: >>> A minor change (also called "cosmetic") usually is a typo fix, doc >>> improvement, a little code refactoring that don't change the behavior etc. >>> >>> In Wikipedia we can mark an edition as "minor". >>> >>> It would be nice to have an argument like "--minor" in git-commit to >>> mark the commit as minor. Also, filter in git-log (like --hide-minor) to >>> hide the minor changes. The git-log could be optimized to show minor >>> commits more discreetly. >>> >>> >> >> This should just be part of the commit message log, generally projects >> use something like TRIVIAL in the patch subject or similar. You could >> also standardize for your project(s) what would be considered a minor >> change. The issue is that not everyone considers these changes as >> "minor". You should be able to use a combination of the --grep option >> in log to search for all commits who don't contain that string in the >> right format. > > Could also be a good use for notes, since you might want to add this > markup after the fact. >