On Sat, May 07, 2005 at 11:24:16PM +0200, Michael Banck wrote: > On Mon, May 02, 2005 at 11:01:57AM +1000, Matthew Palmer wrote: > > And all of these problems can be averted by doing the right thing, and > > using a real revision control system and then shipping a regular ol' > > "everyone knows how this works" debhelper-based .diff.gz. > > Eh, and how does this help QA? If I see a .diff.gz with 20 patches > nicely intertwined into each other, I just rm -r the source tree and go > for the next package. Unless I really have to hack on it, at which > point I shake my fist in anger. Because you see the package as it will be built, and can just dive in with your debugger or other appropriate tool and go to work fixing the symptom you're trying to fix, without having to work out Yet Another weird-arse patch management/build system. The only situation I can see, from a QA perspective, where having each patch split out is good, is when you suspect that a particular maintainer-applied patch is terminally broken. But when you're attacking a package for the first time, you will rarely know what's upstream and what's maintainer-applied, so I don't see the big difference -- you squash the bug and you move on. - Matt
Attachment:
signature.asc
Description: Digital signature