[Pkg-haskell-maintainers] Bug#542842: HSH only on arches with threaded RTS?

[Kurt Roeckx]

On Sat, Aug 22, 2009 at 01:36:21AM +0200, Joachim Breitner wrote:
> Hi,
> 
> Am Freitag, den 21.08.2009, 14:34 -0500 schrieb John Goerzen:
> > On Fri, Aug 21, 2009 at 06:49:26PM +0000, Joachim Breitner wrote:
> > > Hi John,
> > > 
> > > I have hit a similar problem with hlint:
> > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=541679
> > > 
> > > I guess you???ll have to make the -parallel parameter depend on a cabal
> > > flag, and that it based on the arch you are compiling on, or dynamically
> > > based on the existence of the threaded runtime.
> > 
> > Thank you for the hint!  I use -threaded (not -parallel) in the
> > .cabal, but I did notice a hackage warning recently saying that
> > -threaded has no impact on libraries.  Aside from the tests, HSH is
> > library only.  I'll see about just removing -threaded entirely.
> > 
> > [ later ]
> > 
> > Oh drat, now I'm confused.  The problem was when it tried to run
> > tests.  Does this mean that sparc64 doesn't support threaded RTS?  HSH
> > does now require threaded RTS, I believe.
> 
> yes, some arches just do not have the threaded RTS. If HSH does not work
> there, then that's the way it is. Although I'm not sure what's the
> correct step then (Set the correct arches in the Arch: line, or asking
> for an entry in the packages-arch-specific file. Probably the latter,
> I'm CC'ing the wanna-build team as they might know.)

I suggest you start by changing it so that it fails early in the
built process and that it's clear that the arch isn't supported
(yet).  Is there something you can run at built time to test it?
I would recommend such a test in any case.

Restricting the arch list will probably make it fail early, and
should work just as well.

We can also add it to Packages-arch-specific to stop attempting
to built on the arches that aren't supported.


Kurt