Development stage timeline?

Development stage timeline?

[Damian Rouson]

All,

Could anyone provide an estimate of when GCC development will move to the next stage of development?  I assume that date is the deadline for submitting patches that support new features.

________________________________
Damian Rouson, Ph.D., P.E.
President, Sourcery Institute
http://www.sourceryinstitute.org
+1-510-600-2992 (mobile)

Re: Development stage timeline?

[Janne Blomqvist]

On Wed, Oct 28, 2015 at 7:08 PM, Damian Rouson
<damian@sourceryinstitute.org> wrote:
> All,
>
> Could anyone provide an estimate of when GCC development will move to the next stage of development?  I assume that date is the deadline for submitting patches that support new features.

You can find the latest status report here:
https://gcc.gnu.org/ml/gcc/2015-10/msg00113.html

(In principle, AFAIU, Fortran isn't bound by this as tightly as the C
and C++ frontends, so the Fortran maintainers have some leeway to
accept patches at their discretion. That being said, the schedule is
there for a good reason, so it might make sense to follow it
nonetheless.)


-- 
Janne Blomqvist

Re: Development stage timeline?

[Paul Richard Thomas]

Dear Damian and Janne,

I have beaten my head against various approaches to parameterized
derived types and have come up against the proverbial brick-wall. I
can see how to do it now, having tried so many blind alleys. Whether
or not I can come up with something before the end of stage-1, I don't
know. By way of light relief, I have been clearing my backlog of PRs
:-)

The trick with pdts is going to be to turn them into a set of macros,
that substitute the kind parameters and produce appropriate
initialization procedures, using the len parameters, and finalization
routines. I have been researching various ways of doing this,
including m4, but have concluded that rolling my own is the best way
to proceed.

As Janne says, we are not so tightly bound by the deadlines as C and
C++ and so there is a little bit of leeway.....

I wonder if Jerry has made any progress with derived type IO?

Cheers

Paul

On 28 October 2015 at 18:15, Janne Blomqvist <blomqvist.janne@gmail.com> wrote:
> On Wed, Oct 28, 2015 at 7:08 PM, Damian Rouson
> <damian@sourceryinstitute.org> wrote:
>> All,
>>
>> Could anyone provide an estimate of when GCC development will move to the next stage of development?  I assume that date is the deadline for submitting patches that support new features.
>
> You can find the latest status report here:
> https://gcc.gnu.org/ml/gcc/2015-10/msg00113.html
>
> (In principle, AFAIU, Fortran isn't bound by this as tightly as the C
> and C++ frontends, so the Fortran maintainers have some leeway to
> accept patches at their discretion. That being said, the schedule is
> there for a good reason, so it might make sense to follow it
> nonetheless.)
>
>
> --
> Janne Blomqvist



-- 
Outside of a dog, a book is a man's best friend. Inside of a dog it's
too dark to read.

Groucho Marx

Re: Development stage timeline?

[Jerry DeLisle]

On 10/28/2015 12:41 PM, Paul Richard Thomas wrote:
... snip ...

> I wonder if Jerry has made any progress with derived type IO?
> 

Unfortunately no, however I happened to start looking at it some more today.

Jerry

Re: Development stage timeline?

[Damian Rouson]

> On Oct 28, 2015, at 10:15 AM, Janne Blomqvist <blomqvist.janne@gmail.com> wrote:
> 
> On Wed, Oct 28, 2015 at 7:08 PM, Damian Rouson
> <damian@sourceryinstitute.org> wrote:
>> All,
>> 
>> Could anyone provide an estimate of when GCC development will move to the next stage of development?  I assume that date is the deadline for submitting patches that support new features.
> 
> You can find the latest status report here:
> https://gcc.gnu.org/ml/gcc/2015-10/msg00113.html
> 
> (In principle, AFAIU, Fortran isn't bound by this as tightly as the C
> and C++ frontends, so the Fortran maintainers have some leeway to
> accept patches at their discretion. That being said, the schedule is
> there for a good reason, so it might make sense to follow it
> nonetheless.)
> 
> 

Thanks for this information.  I’m hoping Alessandro (cc’d) will review and submit the Fortran 2015 EVENT patch that Tobias originally posted to this list in April:

https://gcc.gnu.org/ml/fortran/2015-04/msg00119.html

Given that Tobias’s email states the following, I hope this is a strong candidate for an exception to the stage rules if necessary:

 "I think the patch is finished, but I still have to extend the test cases, to re-read the patch and to write a changlog."

Alessandro submits his dissertation tomorrow (hooray!) and expects to resume work on gfortran/OpenCoarrays soon thereafter.   EVENT support already exists in OpenCoarrays so we are just waiting for support on the compiler side.  We are optimistic that judicious use of the EVENT features can have a positive impact on parallel performance as compared to SYNC ALL and SYNC IMAGES.

One important step Alessandro will need to take is to ensure that any recent changes to the relevant TS are reflected in the patch. It’s possibly there is nothing to do in that regard, but the good news is that the TS passed a straw ballot last month and is being forwarded to ISO for approval and publication so I think it’s safe to say that it is essentially final now.  That makes this a very opportune time to get the feature into the compiler.  

I don’t think anyone responded to Tobias’s April email.  I hope there is someone who is willing to review the patch when Alessandro or Tobias submits a candidate final version.

Damian

Re: Development stage timeline?

[Damian Rouson]

> On Oct 29, 2015, at 10:45 AM, Jerry DeLisle <jvdelisle@charter.net> wrote:
> 
> On 10/28/2015 12:41 PM, Paul Richard Thomas wrote:
> ... snip ...
> 
>> I wonder if Jerry has made any progress with derived type IO?
>> 
> 
> Unfortunately no, however I happened to start looking at it some more today.

Bravo!  I’m cheering you from the sidelines.  I truly believe that the day that the DTIO And PDT patches are in the trunk will be a watershed moment.  Presumably gfortran will then be very close to full Fortran 2008 compliance.  I’m betting there at that point be there fully Fortran 2008 compliant compilers (Cray, Intel, and gfortran), one nearly Fortran 2008 compiler (IBM), and two Fortran 2003 compliant compilers (NAG and Portland Group).  I will be doing a happy dance when this day arrives.

:D 

Re: Development stage timeline?

[Damian Rouson]

> On Oct 28, 2015, at 12:41 PM, Paul Richard Thomas <paul.richard.thomas@gmail.com> wrote:
> 
> Dear Damian and Janne,
> 
> I have beaten my head against various approaches to parameterized
> derived types and have come up against the proverbial brick-wall. I
> can see how to do it now, having tried so many blind alleys. Whether
> or not I can come up with something before the end of stage-1, I don't
> know. By way of light relief, I have been clearing my backlog of PRs
> :-)
> 
> The trick with pdts is going to be to turn them into a set of macros,
> that substitute the kind parameters and produce appropriate
> initialization procedures, using the len parameters, and finalization
> routines. I have been researching various ways of doing this,
> including m4, but have concluded that rolling my own is the best way
> to proceed.

It’s great to hear that you can see light at the end of the tunnel — or at least a pathway to the light.

:D