The "smoke testing" has completed.
While {Free,Net,Open}BSD were a mess, the following worked fine with Jeff's
tarball:
Mac OS X 10.6, 10.7 and 10.8 on x86-64 (LP64 and ILP32 ABIs)
Solaris-10 on SPARC (v8+ and v9 ABIs)
Solaris-11 on X86-64 (LP64 and ILP32 ABIs)
The *BSD platforms when
On Tue, Feb 24, 2015 at 1:45 PM, Paul Hargrove wrote:
[...]
>
> Smoke testing will begin momentarily...
>
[...]
I am choking on all the smoke.
Somebody call the fire marshall!
It looks like with Jeff's tarball all the BSDs are failing in the same way:
---
Dave --
That's an unfortunate segv in the runtime, not the the MPI layer. Did you get
a corefile, perchance? Could you send a backtrace?
> On Feb 23, 2015, at 9:53 PM, Dave Turner wrote:
>
> Jeff, George,
>
> When I try to use yalla with the dev master I get the error message
> belo
On Feb 24, 2015, at 4:45 PM, Paul Hargrove wrote:
>
> Check for dlfcn.h and the dlopen symbol in -ldl.
>
> Then the paranoid part of me wants to note that since you don't try using
> dlopen() in the configure tests you risk encountering platforms with
> non-functional/non-conforming implementa
See two responses inline below.
On Tue, Feb 24, 2015 at 1:08 PM, Jeff Squyres (jsquyres) wrote:
> On Feb 24, 2015, at 1:55 PM, Paul Hargrove wrote:
> >
> > Forgive me for asking a question I am sure I could answer by reading the
> .m4:
> > How are you planning to distinguish which platforms su
On Feb 24, 2015, at 1:55 PM, Paul Hargrove wrote:
>
> Forgive me for asking a question I am sure I could answer by reading the .m4:
> How are you planning to distinguish which platforms support dlopen()?
Check for dlfcn.h and the dlopen symbol in -ldl.
> And the question you should have seen co
Jeff,
+0.95
Read the new PR yesterday and agree it makes sense to bypass libltdl where
it would add little or nothing to a "dlopen-lovin' platform".
Forgive me for asking a question I am sure I could answer by reading the
.m4:
How are you planning to distinguish which platforms support dlopen()?
Short version
=
I think I have a PR that now solves the libltdl issue. See
https://github.com/open-mpi/ompi/pull/410 if you care.
If not one has any objections, I'll merge this tomorrow (Wed 25 Feb 2015).
More detail
===
Original problem (can't upgrade Libtool beyond 2.4.2
Dave,
The PML is the layer with the smartness (or so we hope) that moves the data
around in Open MPI. The default PML is called OB1, and is the only one that
supports multiplexing over multiple transports. Under certain
circumstances, other, more specialized PMLs (such as yalla or mxm) take
preced
