Ola!
I was trying to re-re-fix https://ghc.haskell.org/trac/ghc/ticket/8089 and add a test-case for this fix. However, I'm having trouble creating a test that can actually be successful. The simplest case exhibiting the issue (on OSX) is that "main = threadDelay maxBound" crashes, I need a way to treat a timeout of say 5-10s as success, since the testsuite won't let my test run for 25,000 days :D
I can't add a timeout inside the haskell code (along the lines of "threadDelay 10000000 >> exitSuccess" since installing any other timers with a timeout shorter than the crashing one masks the crash. Marking the test as expected to fail doesn't help, as this treats both a crash due to the bug and a timeout as success. Clearly only timeout should be considered successful.
Cheers,
Merijn
PS - For the curious willing to help verify that the third time's the charm: https://phabricator.haskell.org/D407
I think this is a great idea. My understanding from the wiki page is that
the full plan involves:
1. We need to *fix compatibility with recent LLVM versions*. This really
sucks for users. *Ben Gamari* is working on this, see #9142
<https://ghc.haskell.org/trac/ghc/ticket/9142> and ​Phab:D155
<https://phabricator.haskell.org/D155>
2. We need to fix up the LLVM IR generation in GHC, and fix the compiler
driver to invoke the tools with a better set of optimizations. *Nathan
Howell* has been wrangled into looking into this.
3. We need to ship a pre-built version of LLVM for developers and users,
and put them in binary distributions for major platforms. *Austin Seipp*
could handle this.
4. We can fix other bugs, like LLVM -split-obj support, which opens the
way to shipping GHC with LLVM-optimized base libraries. *Reid Barton*
expressed some interest in this.
I assume that 1. will come with 7.10.1. Is there any way we could do 3. in
one of the 7.10 releases before 7.12.1?
Basing this question on Austin's words "we *only* need opt and llc, so the
distributions can be tiny. " and "it's not actually very complicated to
implement or support, I think." :) I guess it might be more difficult than
I think but I thought it was worth asking the question as I believe 3. on
its own is worthwhile and would hopefully simplify support for LLVM in GHC.
Thanks
George
On Fri, Oct 24, 2014 at 8:52 PM, Austin Seipp <austin(a)well-typed.com> wrote:
> Hi *,
>
> A few days ago a discussion on IRC occurred about the LLVM backend,
> its current status, and what we could do to make it a rock solid part
> of GHC for all our users.
>
> Needless to say, the situation right now isn't so hot: we have no
> commitment to version support, two major versions are busted, others
> are seriously buggy, and yet there are lots of things we could improve
> on.
>
> So I give you a proposal, from a few of us to you all, about improving it:
>
> https://ghc.haskell.org/trac/ghc/wiki/ImprovedLLVMBackend
>
> I won't repeat what's on the wiki page too much, but the TL;DR version
> is: we should start packaging a version of LLVM, and shipping it with
> e.g. binary distributions of GHC. It's just a lot better for everyone.
>
> I know we're normally fairly hesitant about things like this (shipping
> external dependencies), but I think it's the only sane thing to do
> here, and the situation is fairly unique in that it's not actually
> very complicated to implement or support, I think.
>
> We'd like to do this for 7.12. I've also wrangled some people to help.
> Those people know who they are (because they're CC'd), and I will now
> badger them into submission until it is fixed for 7.12.
>
> Please let me know what you think.
>
> PS. Joachim, I would be particularly interested in upstream needs for
> Debian, as I know of their standard packaging policy to not duplicate
> things.
>
> --
> Regards,
>
> Austin Seipp, Haskell Consultant
> Well-Typed LLP, http://www.well-typed.com/
> _______________________________________________
> ghc-devs mailing list
> ghc-devs(a)haskell.org
> http://www.haskell.org/mailman/listinfo/ghc-devs
>
