- From: Terje Bless <link@pobox.com>
- Date: Fri, 29 Nov 2002 12:32:18 +0100
- To: QA Dev <public-qa-dev@w3.org>
- cc: Ville Skyttä <ville.skytta@iki.fi>
Wooops! That was supposed to go to QA Dev. Serves me right for answering email after 24 hours without sleep. :-) Terje Bless <link@pobox.com> wrote: >Ville Skyttä <ville.skytta@iki.fi> wrote: > >>I'm confused about where to commit validator changes at the moment, to >>0_6_0-branch or HEAD? Are you planning a 0_6_0 -> HEAD merge after >>0.6.1 is out? > >Short version: check in to 0_6_0-branch or HEAD depending on whether >it's small fix that should be in the 0.6.1 release or a larger change >that is targetted for "some future version". Yes, I plan to merge 0.6.x >onto the trunk after 0.6.1 is out the door. > > >Long Version: > >I'm planning a 0.6.1 release (hopefully even some time around this >weekend) with mostly just bug fixes. This is what's happening on the >validator-0_6_0-branch. > >HEAD is a freeforall (more or less) to add big or distruptive new stuff. >Eventually we'll branch for validator-0_7_0-(branch|release) and any >maintenance releases for that. > >Once 0.6.1 is out the door I'll merge validator-0_6_0-branch onto HEAD. >Similarly if we need a 0.6.2 release; including a merge after that too. > > >The idea is that HEAD is at any given moment almost by definition broken >and in flux. When we want to work towards a release we split off a >branch where we can be as anal as necessary about changes to achieve >stability without hindering new feature work on HEAD. > >Merging after a release, instead of doing it concurrently, is just so >small fixes to a release branch don't have to be slowed down by >determining how and if they shoudl be applied to HEAD. Especially since >many bug fixes -- both in terms of what they fix and in how they are >implemented -- are specific to the release branch and/or no longer apply >to the HEAD. > > >For /really/ disruptive stuff I envision making experimental branches >even off from HEAD so that you can still work on HEAD while someone is >off redefining the world (typically; the template system or modularizing >the validator). > >It's a bit more expensive because you have to spend time merging >periodically, but I think it'll end up being much cleaner in the long >run. -- "I don't mind being thought of as a badguy, but it /really/ annoys me to be thought of as an *incompetent* badguy!" -- John Moreno
Received on Friday, 29 November 2002 06:32:39 UTC