On 06/06/2011 16:49, Steffen Schuldenzucker wrote: > On 06/06/2011 03:50 PM, Miles Gould wrote: lead to a consistent state again. >> >> I think it's worth asking how much of an actual *problem* you've been >> finding this. Do you often find yourself pulling patches and getting a >> non-compiling source tree because of some missing dependency? How much >> time have you wasted because of this issue? > > It is a theoretical question. As I am writing my thesis alone and this > is a pretty linear process, I haven't really been hit by the issue at all. > But I imagined that, in a larger project with many collaborators, it > gets annoying quickly. Which seems to be wrong.
I can think of of a couple of main cases where I try to cherry-pick: (a) I'm pushing to upstream/sending in a patch, and I want to just send in one change of the many I'm currently working on. In that case I check before doing the push, either by testing in a staging repo or by eye. If it turns out there's an accidental dependency - either detected by darcs or not - I amend to get rid of it. (b) I'm pulling a specific change from upstream In this case I treat cherry-picking as opportunistic; if it doesn't work (and I check by building etc) then I just live with it. darcs makes cherry-picking easy, and crucially it makes it easy to re-merge with upstream after cherry-picking. But that doesn't mean it's always possible, and the under-approximation does place the onus on the user to check. Cheers, Ganesh _______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
