Comment # 9 on bug 1172723 from Stefan Br�ns

(In reply to Atri Bhattacharya from comment #8)
> (In reply to Stefan Br�ns from comment #7)
> > Although Atri does a good job in general, the VTK 9.0 update was IMHO rushed
> > (and thats the reason I did not accept the SR, unfortunately someone else
> > did ...):
> > 
> > 1- Fails to build on ARM
> 
> A patch would be very helpful, but it would need to be either accepted or
> acceptable upstream. I would like to avoid carrying patches that require
> extensive rebasing over and over.
> 
> Btw, this isn't only about this patch: vtk and paraview had been carrying
> multiple patches for a year before I went and poked upstream about them (a
> look at the changelog should make it clear what I am talking about).

One of these patches being for proj4, which I had reported and submitted
upstream, https://gitlab.kitware.com/vtk/vtk/-/issues/17554

The GLES patch was also submitted upstream, but ignored for more than a year.
I have just rebased it, hopefully it gets included this time.

I also poked gl2ps upstream to do a new release, so the package no longer
relies on the bundled gl2ps.

So, as you can see you are not the only one trying to push things upstream.
But sometimes I lack the time to immediately do everything at once ...

> > 2. Java bindings are removed on 15.1, but devel dependencies still reference
> > Java package
> > 
> 
> Yes, sorry about that. My first instinct was to require cmake >= 3.12 and
> make it unresolvable for Leap 15.1 which probably would have been better.
> 
> In any case, I would appreciate if you actually __decline__ the sr the next
> time... You had ample time and opportunity to do so for this.

I did a first review, and after that I was busy. Unfortunately I trusted you
and Dirk to check if there is any breakage. So next time I am busy I obviously
have to reject the SR because it *may* contain errors ...

> > and most importantly:
> > 3. VTK 9.0 is API incompatible to FreeCAD and CSXCAD, the two users of VTK
> > in TW.
> 
> Surely, that's for these apps to catch up. Thankfully, this is not one of
> the policies for TW; otherwise, it will make updating any package quite
> difficult.

Thats obviously wrong. FreeCAD and CSXCAD are not part of the Staging projects,
otherwise the update would have been rejected automatically. For packages not
in Staging, it is the responsibility of the submitter to check if the update
does not cause any major breakage.

Honestly, what use is a update which breaks *all* users of VTK in TW? Even more
so as the VTK update coinceded with the GCC10 and Qt5.15 update, and now there
is a FreeCAD in the repo which can't be installed (requires Qt5.14 python
bindings) and fails to build.