[Llvm-bgq-discuss] linker errors with llvm/current/bin/bgclang++

[Hal Finkel]

Geoffrey,

I've installed the rebased build in /home/projects/llvm/r174210-20130201; I've not yet switched it to the current default (so you'd need to set MPICH_CC and/or MPICH_CXX to use it with the mpi wrapper scripts) because I'm investigating some performance regressions. Nevertheless, I'd like to know whether or not this fixes your other problems.

Also, hopefully you won't need -mcmodel=small with this build; if you still need it then please let me know that as well.

Thanks again,
Hal

----- Original Message -----
> From: "Geoffrey Irving" <irving at naml.us>
> To: "Hal Finkel" <hfinkel at anl.gov>
> Cc: llvm-bgq-discuss at lists.alcf.anl.gov
> Sent: Thursday, January 31, 2013 3:33:57 PM
> Subject: Re: [Llvm-bgq-discuss] linker errors with llvm/current/bin/bgclang++
> 
> On Sun, Jan 27, 2013 at 9:48 PM, Hal Finkel <hfinkel at anl.gov> wrote:
> > ----- Original Message -----
> >> From: "Hal Finkel" <hfinkel at anl.gov>
> >> To: "Geoffrey Irving" <irving at naml.us>
> >> Cc: llvm-bgq-discuss at lists.alcf.anl.gov
> >> Sent: Thursday, January 24, 2013 8:52:42 AM
> >> Subject: Re: [Llvm-bgq-discuss] linker errors with
> >> llvm/current/bin/bgclang++
> >>
> >> ----- Original Message -----
> >> > From: "Geoffrey Irving" <irving at naml.us>
> >> > To: "Hal Finkel" <hfinkel at anl.gov>
> >> > Cc: llvm-bgq-discuss at lists.alcf.anl.gov
> >> > Sent: Wednesday, January 23, 2013 3:52:07 PM
> >> > Subject: Re: [Llvm-bgq-discuss] linker errors with
> >> > llvm/current/bin/bgclang++
> >> >
> >> > I'm still seeing this error.  If you've already rebased, should
> >> > I
> >> > file
> >> > a bug upstream?
> >>
> >> I've not yet rebased. I hope to get to that today.
> >
> > There was a bunch of prep work I had to do prior to rebasing, and I
> > think that's now done. I'll start rebasing tomorrow, and I hope
> > that it will be done by Tuesday. I'll send out a general e-mail
> > when the new version is ready for testing.
> 
> No worries about the delay: I have a workaround for the compiler
> crash bug:
> 
>     https://github.com/girving/pentago/commit/551a78502f323a3941219339036da10403d1c72a
> 
> The problem is the following overload mixing variadic with
> nonvariadic
> templates:
> 
>     https://github.com/otherlab/core/blob/da0965dee8e6c03f66d5a3e54ec5b0a7f3f302ee/vector/Vector.h#L526
> 
> I only used this in a few places in my pentago code, so it was easy
> enough to call the constructor directly.
> 
> Geoffrey
>