On Tuesday 04 December 2007, Joey Hess wrote: > Frans Pop wrote: > > I've checked the changelog, but did not see anything to indicate this > > is an intended behavior change. It also seems illogical to me. > > * debchange: Default to using -t / --mainttrailer when using the > "changelog" heuristic. Thanks to Joey Hess for the patch
Hmm. Missed that. Was looking for dch, not debchange. > This avoids a majority of merge conflicts, I don't see anything illogical > about it since the sig is updated on dch -r. Because the sig no longer matches the last update. I can kind of understand the reasoning behind the change, but if you want the sig to be meaningless for UNRELEASED changelog entries, it should IMO be really meaningless, i.e. containing a fictional name/email and a clearly incorrect date, like: -- Anonymous <[EMAIL PROTECTED]> Sat, 01 Jan 0000 00:00:00 +0000 Keeping the sig/date for the random first person who opens a new release entry makes no sense to me. In that case I expect the sig to be updated when I modify the changelog. [1] date tells me that Sat is the correct day for 01-01-0000 :-) $ date -Rud "01 Jan 0000" Sat, 01 Jan 0000 00:00:00 +0000
signature.asc
Description: This is a digitally signed message part.
