[Debian-med-packaging] Bug#753809: Re: Bug#753809: ginkgocadx: will not display studies

[Karsten Hilbert]

Hello Mathieu,

thanks for continuing to have a useful discussion.

> You found a bug in A (ginkgocadx),

That is correct - in a way. I found a bug when using A on
Debian. I then went upstream and used their static binary of
A compiled for Debian. That binary worked. This meant that
the bug lies in "the Debian variant of A".

Nevertheless I reported the bug to A-upstream because just
because it doesn't happen in upstream doesn't mean it doesn't
_exist_ in upstream.

> which uses B (ITK) which in turns uses C (GDCM).
> Upstream for A says: "this is not our fault it is somewhere in B or C".

I agree with you here. I would assume upstream decided that
they don't want to afford searching for a bug in a
non-upstream variant of their code. Good or bad -- it is
their decision.

> I believe somewhere in between A or B there is a missing link to
> report *exactly* what is wrong with C.

You are correct. That missing link would probably be me
because I am a user who wants to use A but in conjunction
with a version of B and/or C which A-upstream doesn't
directly support. However, I lack the skills to nail the
problem myself.

Therefore I went A-upstream and asked for details. I then
reported what I got as a response. I agree this response
would be a lot more helpful if A-upstream pointed out the
exact issues they had with in-Debian B and/or C.

> In summary yes I expect a bug report for C to be somewhat minimalist,
> but don't ask me to use A -> B -> C to dig a bug somewhere in A. The
> chain as I see it is that either someone from A or B is capable of
> reporting a reduced test case for C.

So, in effect, A (or me equipped with information gotten from
A) should report a bug against B (ITK). Then B (or me,
equipped with information gotten from B) would report a bug
against C (GDCM).

Or else A would provide information on the exact problem
against C in which case B can be skipped.

Now, since A doesn't seem to afford the resources to help us
find bugs in our use of their code I guess we are stuck.

In short: Debian doesn't have a DICOM viewer suitable for
clinical care.

Which is a great pity.

BTW: I do acknowledge that Dmitry has at least tried to
reproduce the problem.

BTWBTW: I also know that all of you guys volunteer your time.

Karsten
-- 
GPG key ID E4071346 @ eu.pool.sks-keyservers.net
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346