[kgraphviewer-devel] KF5 version of kgraphviewer

[Milian Wolff]

On Monday 04 August 2014 22:42:24 Reimar Döffinger wrote:
> On 04.08.2014, at 22:34, Arnold Dumas <contact at arnolddumas.fr> wrote:
> > Le 04/08/2014 22:17, Reimar Döffinger a écrit :
> >> Hello!
> >> 
> >> On Mon, Aug 04, 2014 at 10:06:54PM +0200, Arnold Dumas wrote:
> >>> Massif-visualizer is being ported to KF5. As you might know,
> >>> massif-visualizer uses kgraphviewer to show nices graphs.
> >>> At the moment, there is no KF5-based version of kgraphviewer. I'm then
> >>> asking you: is there anybody working on such a port. If no, I'll be glad
> >>> to
> >>> work on it.
> >> 
> >> I don't think anyone is, I think we're kind of challenged time-wise
> >> just doing the basic maintenance :( .
> >> And I don't even have any idea what kind of changes that means.
> >> If it means any changes that aren't backwards-compatible I expect
> >> we will want to wait with committing anything to trunk until we made
> >> a proper release - though with git using a branch admittedly shouldn't
> >> be an issue really.
> > 
> > Basically a KF5-based version means, based on Qt5 and it's not
> > backward-compatible.
> That is unfortunate, the few articles I skimmed about it sounded like at
> least the first pass should be perfectly possible in a way that works both
> with Qt4 and Qt5.

It is possible but with no gain (imo). Let's just release kgraphviewer one 
last time for KDE4 (the 2.2 branch is there for that). A frameworks branch 
could then eventually be merged into master and we'll phase out kde4 support.

> > The common practice is to create a "frameworks" branch and then to do the
> > stuff there. The idea isn't to break master nor to mess up existing
> > working versions, it's to make this application moving forward to the KDE
> > future, aka Frameworks 5.
>
> It just means that until we are ready to drop Qt4 support someone has to
> maintain that branch. If you think you have the time for that, great.
> Otherwise having as much as reasonably possible on trunk/master reduces the
> risk of losing work ("losing" as in "merging the branch to trunk would be
> more effort than just re-implementing the changes").

Considering that there is virtually no maintenance of the KDE4 branch anyways, 
I guess porting it to KF5 and just concentrating on that wouldn't hurt either. 
I mean most other projects are going that way after all. But generally, 
merging a legacy KDE4 branch into a KF5 branch should be possible without too 
many issues, as long as only the bare minimum porting is done. As soon as the 
whole codebase is touched with cleanups (which are really required in 
KGraphViewer land, unrelated to KF5 or KDE4), we'll get into merge hell - but 
again, unrelated to KF5.

So personally, I'd say lets start a KF5 branch.

Bye

-- 
Milian Wolff
mail at milianw.de
http://milianw.de