next snapshot (Re: kdesupport/qt-dbus)

[Matt Rogers]

On Sunday 18 June 2006 18:01, Matt Rogers wrote:
> On Sunday 18 June 2006 15:43, David Faure wrote:
> > On Sunday 18 June 2006 16:37, Thiago Macieira wrote:
> > > SVN commit 552611 by thiago:
> > >
> > > Introduce the new QtDBus type marshaller/demarshaller system, dubbed
> > > "Pattern Buffer". This change is binary-incompatible, but
> > > source-compatible. If you can't afford to rebuild everything, please
> > > wait until the next kdelibs snapshot (which should happen tomorrow).
> >
> > OK. Actually I said last week "we should wait until all main modules are
> > ported to dbus before making more snapshot updates", but if there are
> > changes that actually help with the porting, then that's a reason to
> > update the snapshot indeed. Well, technically those are changes in
> > qtdbus, not in kdelibs, but while we have to recompile kdelibs, we might
> > as well update it if it doesn't bring in more API changes to port to.
>
> I will argue that the only thing that needs to be updated is the relevant
> revisions that are related to the dbus changes. Updating the snapshot more
> often only creates problems for those who expect to have a decently stable
> base to port against, since that's what the snapshot is for.
>
> The recreation of the snapshot last monday completely violated that
> expectation. We might have well have been using trunk in that case.
>
> > I had a look at what would change if updating kdelibs snapshot, and only
> > this came up: - many bugfixes (including knotify and kded working better,
> > KPageWidget stuff, the kdatetime static-object crash fixed etc.) - the
> > removal of the name argument in the KDEDModule class
> > - the new KButtonGroup class
> > - ah and maybe the removal of q3 includes might mean some more q3
> > includes to add in modules. No big change, so it shouldn't harm the
> > current porting effort.
>
> these are the only things that come up because you updated the snapshot
> last week. Otherwise, the list would be longer.
>
> > => I suggest that we update the snapshot tomorrow indeed.
> > But after that I suggest that we stick to "we'll only update the snapshot
> > when all main modules have been ported to dbus", otherwise we'll never
> > have compiling modules.
>
> tomorrow is the normal day to update the snapshot on the every two week
> update schedule. When you updated it last week, you were a week ahead of
> schedule, and IMO, it should have never been updated then (or at the very
> least, announced on kde-cvs-announce first)
>
> I say stick with the current schedule, and backport important revisions on
> an as needed basis.

just so it's clear, I'm not upset about the snapshot being updated. What I am 
upset about is the lack of an announcement or some other kind of warning 
(and, of course, I may have missed it) that would have allowed me to prepare 
for the pain. All I'm really asking for is that better attempts are made to 
communicate and manage expectations so that we know things will be broken 
(w/o having to throughly read kde-commits). IMHO, this is what 
kde-cvs-announce is for.
--
Matt