gfortran IRC meeting on Sunday, 20 November

[Janus Weil]

Hi guys,

> Great!  Who is going to bring the beer?
>
> I'll mull over possible points for discussion but, in the mean time,
> count me in.

I'm afraid I'll not be able to make it on Sunday (despite the exciting
prospect of USB beer).

Also, I'm packed up to my ears in physics research right now, which is
why I'm being so silent lately (still alive, though ...)

Anyway, thanks for the great progress with constructors and class arrays ... !

Cheers,
Janus



> On Thu, Nov 17, 2011 at 10:49 AM, Tobias Burnus <burnus@net-b.de> wrote:
>> Hi all,
>>
>> as suggested at  [1], I would like to have a post-Stage-1 meeting in IRC
>> (irc.oftc.net's #gfortran) next Sunday at the usual time: 21:00 (9pm) CET,
>> 20:00 (8pm) UTC, 12:00 (noon) PST.
>>
>> Suggested topics:
>> - Stage 3 work
>> - 4.8 work
>> - miscellaneous
>>
>> For Stage 3: Besides the committal of Paul's almost ready polymorphic array
>> patch, I would like to see some more OOP fixes (including polymorphic
>> coarrays, fixes for issues related to polymorphic arrays and possibly also
>> some of the poly. operator bugs). I also would like to see fixes to the
>> deferred-length support for character variables; this 4.6 feature has still
>> some loose ends, which hamper the use. For coarrays, fixing intrinsic
>> assignments (no reallocation of coarray components) and deallocation support
>> with -fcoarray=lib are on my wishlist. And, of course, we should reduce the
>> number of regressions* - and fix many of the small bugs. (* see [1] for an
>> almost up-to-date list.)
>>
>> For 4.8 we have a huge number of projects which we could work on. For
>> instance: IEEE, true multi-image coarrays, unlimited polymorphism,
>> finalization subroutines. But also: Parameterized derived types /
>> deferred-length character components, user-defined I/O, submodules, and
>> plenty of minor items. Other items like cpp (target macros, nontraditional
>> mode), better diagnostics, additional !GCC$ attributes support, etc. [There
>> was also once the plan to include a POSIX library/module to create
>> directories etc. and to include the ISO Varying strings, though the latter
>> is probably obsolete by now.]
>> And we have the array-descriptor update which is related to the TS 29113
>> descriptor, is an issue by itself (correctness, maximal rank), and plays
>> possibly a role for unlimited polymorphism* and for TS29113 items like
>> type(*) and dimension(..). (* The TS 29113's descriptor contains a
>> field/component to denote the data type.) Additionally, the chance of
>> breaking the ABI should be used to do some cleanup.
>>
>> Miscellaneous: Whatever comes up - such as funding/attracting developers,
>> nice-to-know items [3], WG5 news [2].
>>
>> [1] "Fwd: GCC 4.7.0 Status Report (2011-11-08), Stage 1 is over, Stage 3 in
>> effect immediately", http://gcc.gnu.org/ml/fortran/2011-11/msg00093.html
>>
>> [2] WG5/J3:
>> - There is now a draft corrigendum for F2008, cf.
>> ftp://ftp.nag.co.uk/sc22wg5/N1851-N1900/N1893.pdf
>> - TS 29113 is almost ready for the second round of ISO ballots; latest is:
>> ftp://ftp.nag.co.uk/sc22wg5/N1851-N1900/N1888.txt
>> - The next J3/WG5 work item (besides interpretation requests) is the TS for
>> coarrays; cf. ftp://ftp.nag.co.uk/sc22wg5/N1851-N1900/N1888.txt ; only
>> afterwards, the committee will work on the successor of F2008.
>>
>> [3] There was an announcement by PGI/Nvidia/Cray/CAPS for OpenACC [for
>> C/C++/Fortran], which is similar to OpenMP but for accelerators (GPGPU). GCC
>> might consider to also work on supporting it at some point. Those companies
>> are also in the OpenMP WG on accelerators. See
>> http://www.openacc-standard.org/ for details.
>>
>> Tobias
>>
>
>
>
> --
> The knack of flying is learning how to throw yourself at the ground and miss.
>        --Hitchhikers Guide to the Galaxy
>