Send hpx-devel mailing list submissions to
[email protected]
To subscribe or unsubscribe via the World Wide Web, visit
https://mail.cct.lsu.edu/mailman/listinfo/hpx-devel
or, via email, send a message with subject or body 'help' to
[email protected]
You can reach the person managing the list at
[email protected]
When replying, please edit your Subject line so it is more specific
than "Re: Contents of hpx-devel digest..."
Today's Topics:
1. Re: HPX 1.3.0 preparation (Thomas Heller)
----------------------------------------------------------------------
Message: 1
Date: Wed, 10 Apr 2019 22:38:53 +0200
From: Thomas Heller <[email protected]>
Subject: Re: [hpx-devel] HPX 1.3.0 preparation
To: Hartmut Kaiser <[email protected]>,
[email protected]
Message-ID:
<cajcxaey38rtzi+qmzbn+-xw_lq1nh12dbpf4a5tbqcqguyg...@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
On Wed, Apr 10, 2019 at 2:51 PM Hartmut Kaiser <[email protected]> wrote:
>
> Mikael,
>
> Thanks for pushing this forward!
That's great news indeed!
>
> > One thing I left out from the previous email:
> >
> > We'll bump the minimum supported versions of compilers and other
> > dependencies as well for this release. What I had in mind was the
> > following (and there seemed to be some agreement on this):
> >
> > - GCC >= 5
> > - Clang >= 4
> > - Boost >= 1.61.0 (10 supported releases, roughly 3 years)
> >
> > With more frequent compiler releases, we can (and have to) prune old
> > compilers more often. Supporting compilers and dependencies no older than
> > 3 years seems reasonable, and if there's concensus on this I'll write this
> > down somewhere in the documentation.
>
> I fully agree.
>
> > We can also bump the minimum hwloc version if someone feels like it, but
> > it's pretty painless at the moment. I don't think we can go to 2.0 only
> > yet.
>
> I'd agree. Let's leave support for 1.x in place (minimally 1.11.4? I just
> don't remember...)
>
> > With the compiler version bumps we can seriously start thinking about
> > dropping C++11 support in 1.4.0. We'll have to check how well things work
> > with NVCC (and MSVC?) before we finally drop that though.
>
> Not sure if 'dropping C++11' (whatever that means) in favor of requiring
> C++14 would give us any significant benefit. MSVC is not the issue here,
> luckily, anymore. NVCC seems to be the bigger problem nowadays. Intel is
> struggling as well, they break something else every release.
It's 2019 now ;)
IIRC the biggest problems with the intel compiler is mostly related to
the C++ stdlib being used, so it should be mostly fine, I guess.
For nvcc, I guess we have to move to CUDA 10 (or at least 9) as the
minimum version, or strictly require clang.
So, if dropping C++11 means to allow all C++14 construct
unconditionally, I am all for it. The immediate benefits I see there
is the possibility to remove some macros (especially the ones related
to constexpr and lambda captures).
>
> Regards Hartmut
> ---------------
> http://stellar.cct.lsu.edu
> https://github.com/STEllAR-GROUP/hpx
>
> >
> > Mikael
> >
> > ________________________________________
> > From: Simberg Mikael
> > Sent: Thursday, April 04, 2019 9:01 AM
> > To: mailto:[email protected]
> > Subject: HPX 1.3.0 preparation
> > Hi all,
> >
> > I'm starting to prepare for the next release and am aiming for May 1st for
> > the first release candidate. If all goes well the final release will be
> > two weeks later on May 15th.
> >
> > Keep doing what you've been doing so far (i.e. keep opening PRs to master)
> > but start tying up loose ends and help fix bugs that you think can be
> > fixed.
> >
> > Kind regards,
> > Mikael
>
>
> _______________________________________________
> hpx-devel mailing list
> [email protected]
> https://mail.cct.lsu.edu/mailman/listinfo/hpx-devel
------------------------------
_______________________________________________
hpx-devel mailing list
[email protected]
https://mail.cct.lsu.edu/mailman/listinfo/hpx-devel
End of hpx-devel Digest, Vol 48, Issue 7
****************************************
