[Debian-med-packaging] Bug#989296: [RFR] libgdcm-dev/3.0.8-2 (to close #989296)

[Mathieu Malaterre]

Salut Étienne,

On Fri, Jul 30, 2021 at 9:21 PM Étienne Mollier <emollier at emlwks999.eu> wrote:
>
> Hi Mathieu,
>
> Mathieu Malaterre, on 2021-07-30:
> > Hi all,
> >
> > I've reviewed commit 92bee7344b774b45b66185ed17b040f12fe31c43. I've
> > not verified it fixes the OP symptoms, but this is the right fix IMHO.
> >
> > I've also reviewed cddfeab955f486fba72745b66130480dfec1a2b6 and this
> > is not the right fix, sorry. See #711214 for more context.
>
> No worries, Thank you for your review!  :)
>
> So, if I understood correctly #711214 and #989296, then it seems
> to me that, once vtkgdcmsharpglue and vtkgdcmPython are detected
> properly, the remaining messages will be warnings only.  They
> won't impede the build; that is, until the shared objects are
> effectively needed, in which case one should install the
> components independently.  So the fix to d/rules is sufficient,
> no need to touch d/control to address that bug, I guess.

Right ! The point is that on some arch you still want to be able to
install libgdcm-dev without the cil (C# binding is not available
everywhere) or vtk (may require opengl related package which would be
hard to install on headless server).

Debian packages should not be monolithic just because upstream failed
to implement proper *.cmake components packaging.

2cts