On Sun, Sep 11, 2022 at 05:32:31PM -0700, Junio C Hamano wrote: > > I just want to double-check that the plan is to merge this to master as > > noted above. I had thought you would revert jk/unused-annotation and > > that I'd just re-roll it. I'm perfectly happy with either, but just > > didn't want to add more confusion by posting that re-roll. ;) > > Sorry for making a confusing move. The thing is, the first patch in > this two-patch series builds on top of your "UNUSED(var)" thing. > Its patch text depends on "UNUSED(var)" being there, and it explains > why we ended up using the "var UNUSED" syntax over "UNUSED(var)". > It of course is the right thing to do because "UNUSED(var)" was > already in 'next' when it was written. > > We could rewrite it to pretend as if "UNUSED(var)" never happened; > we prefer to keep experiments that turned out to be dead end and we > are unlikely to revisit out of our history. But I think it makes > sense in this case to leave a record in our history that we consider > that "UNUSED(var)" is a superiour implementation that we would have > used and the only reason why we do not use it for now is Coccinelle. > > So, 'next' has the merge of 'jk/unused-annotation' reverted, but when > 'ab/unused-annotation' was merged, the revert was reverted ;-). > When it graduates to 'master', it will pull 'jk/unused-annotation' > along with it and keeps "UNUSED(var)" in our history, but at its > tip, what we end up using will be "var UNUSED". OK, I am happy with that. When I rewrote the initial commit to go to "var UNUSED" instead of "UNUSED(var)", I was going to explain the alternatives in the commit message. But doing it this way is less work for me. ;) I've converted my further annotation patches to "var UNUSED" already, so those will appear that way from the get-go. If we ever go back to UNUSED(var), the initial ones are just a "git revert" away, but the new ones will have to be converted manually. It's not too bad to do it with a clever use of sed/perl, though. -Peff
