Hello Jan, Le Tue, 18 May 2010 15:53:59 +0200, Jan Holesovsky <ke...@suse.cz> a écrit :
> > Hi Charles, > > On Tuesday 18 of May 2010, Charles-H. Schulz wrote: > > > > To bring a bit more of a context here - with the bugs that appear > > > up-stream, but are filed to go-oo bugzilla (we use > > > http://bugzilla.novell.com for go-oo), it is easy - we fix them, > > > commit to go-oo, and either file the patch up-stream, or create a > > > CWS. The concern are the bugs that appear only in go-oo, but are > > > filed to the IZ. > > > > I have two questions here: > > - is the process you describe invariable? (I mean: do you always do > > that?) > > There might be omissions, I am not sure without further looking - we > track the IZ numbers in our patch applying system, anybody is free to > look & ask why we did not file a concrete patch. But generally yes - > this is the process. Okay, thanks. > > > - to me it strikes me as being more complex than just using Go-OO or > > not. Go-OO may or may not be less stable than OOo, but what seems > > to be the real issue imho , is the combination of the following > > factors for Linux distributors or packagers: > > - choice of source: OOo or Go-OO > > The sources are always the same, on go-oo we only mirror the > up-stream sources (+ modify the whitespace a bit, the famous tabs -> > 4 spaces conversion), but other than that, no code changes. > > > - choice of build system: vanilla or ooo-build > > - patches from Go-OO and/or Debian: partial or total integration. > > Patches are the place where all the code changes are located. And there seems to be differences in "behaviour" depending on the choice of your build system. > > > the combination of two or more of these necessary choices for > > maintainers is very much what creates most of the problem, and > > there is no easy answer. The example I usually give here is the > > difference in quality between the Ubuntu build and the *Suse > > build. > > I would be most interested in concrete examples here. The Ubuntu and > openSUSE builds differ just in the patches they apply, the sources > are the same (see above). Well, here's a very simple example: OOo 3.2 has its seach bar and dialog that is simply not localized inside Ubuntu, which is not the case in OpenSuse. > > > Ubuntu is using > > the OOo source, builds it against the ooo-build system and applies > > patches from Debian and Go-OO. > > Just to clarify, go-oo is just another name for ooo-build [ooo-build > is the historical name, and some of us still use that which probably > causes some confusion, sorry for that :-(]. Thanks for the clarification, I had thought the source code has ended up changing somewhat. > > > Suse, on the other hand, uses Go-OO, > > which means its own branch + ooo-build system and its own patches. > > The quality is not as good as OOo, but it is much, much better than > > with Ubuntu. > > Because there are also bugs that exist only in OOo, but not in go-oo, > am not sure what metrics supports the statement that 'go-oo quality > is not as good as OOo'. :-) Let's put it that way: there seem to be different bugs depending on whether you're using OOo or Go-OO. > > > > We would create an alias like go-oo-b...@openoffice.org, and > > > anything that you identify as a go-oo only bug, you'd just > > > reassign to this alias, instead of closing it as 'invalid'. We > > > would take care of the proper assignment of the bug (all the > > > go-oo people have an IZ account too), and its solution. > > > > > > What do you think, please? > > > > That certainly sounds like a good idea. Perhaps a keyword in IZ as > > well like "go-oo specific" would also be useful. I'm wondering what > > others here think about it too. > > This being said, we are not going to solve the need for a direct > > communication on issues between OOo and its linux packagers, so the > > proposal that we worked out with Ubuntu is still on the table which > > is that we open a category for Linux distros, and eventually merge > > the Go-OO IZ with the OOo IZ. Meanwhile your proposal for an alias > > does make real sense. > > No preference about the keyword; I'd think that it does not help that > much, because it is not a 'mandatory' flag, while the 'assigned to' > is. Okay, so let's go with the alias then. Once you have created it, I wonder if André can give you the relevant permissions? best, Charles. > > Thank you, > Kendy > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@qa.openoffice.org > For additional commands, e-mail: dev-h...@qa.openoffice.org > > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@qa.openoffice.org For additional commands, e-mail: dev-h...@qa.openoffice.org
