[pymvpa] request for review OUPS WRONG PATCH

[Valentin Haenel]

* Yaroslav Halchenko <debian at onerussian.com> [090424]:
> 
> On Fri, 24 Apr 2009, Valentin Haenel wrote:
> > indeed, however i still have an unresolved segfault that stops me from doing
> > this. (see previous mail)
> yeah -- I saw you mentioned segfault, but didn't see any details on what
> is actually happening ;) you might like to debug the beast: strace,
> ltrace, gdb and valgrind are your best friends in this case ;)

argh! ;)

> > > as for upstream -- did you mean us? or libsvm?
> > Yeah, i meant you guys. Not sure if i should wait until the patch becomes
> > applied, and then send a new patch, or if i should just interactively rebase the
> > existing one to squash the bug fix, and then send the new version.
> since you didn't yet push your changes anywhere you indeed might simply
> do evil rebases until you think you are done.  It might be worth
> considering using topgit [1] to take care about such feature-branches...
> ?
> 
> [1] http://repo.or.cz/w/topgit.git

I'll have to look at this in detail. Here is my current opinion:

For a couple of lines, rebases are o.k. For anything as large as the searchlight
modifications i did, probably not. (In this case they really are evil!) What i
really want to do is develop features in topic-branches, and then for a patch submission
squash all commits, for example with a squash merge or something like that.

V-