[Pkg-mono-devel] Fw: Bug#255863: muine: still FTBFS because Build-Depends are wrong

Eduard Bloch edi@gmx.de
Mon, 19 Jul 2004 09:30:59 +0200


#include <hallo.h>
* Dave Beckett [Wed, Jul 07 2004, 11:07:56AM]:
> This really seems something outside the scope of Muine.  Although
> amd64 is not a current build architecture, I guess he's saying that
> the cli-virtual-machine (I mispelt it, ignore that) is broke
> as a mechanism for Build-Depends.

cli-virtual-machine is a virtual dependency and should not be used in
Build-Depends

> I think it would be silly and unmaintainable for each mono(*)
> application to have to know which architectures had mono-jit or
> mono-mint running and have to specify them as he outlines.  That
> means maybe going with the virtual package idea.  Maybe mono-cli
> would be better?

What is "mono-cli"? To "solve" the problem, it would need to be a real
package depending on the mono runtime for each arch. However, since
muine seems to be mono-specific and mono-jit seems to be really borken
on any arch except of i386 (and halve-borken on ppc, SMP issues), I do
not see a problem in specifying:

mono-jit [i386 ppc], mono-mint

> Suggestions?  add a wishlist bug on mono and close this?  Add the
> dependencies for now as a short-term fix?
> you changed the Build-Depends from 'mono-jit (>=3D 1.0)' to 'mono-jit
> (>=3D 1.0) | cil-virtual-machine' in an effort to make muine build with

Oh, yes.

> MfG
> 	Goswin

Hallo Goswin, I told you already that I am pissed by how it works now
and it is pretty stupid that the auto-builders break if the first (and
always the first) alternative does not exist. We should be allowed to
specify:

mono-jit | mono-mint

so at least the second one does exist.

Regards,
Eduard.
--=20
Stilbl=FCten aus Schreiben von Versicherungsnehmern:
Ich habe gestern auf dem Heimweg in etwa 20 Meter L=E4nge einen Zaun
umgefahren.  Ich wollte Ihnen den Schaden vorsorglich melden, bezahlen
brauchen Sie nichts, denn ich bin unerkannt entkommen.