Trunk in Stage 3.

Trunk in Stage 3.

[Steve Kargl]

If you missed the announcement, trunk is now in Stage 3.

https://gcc.gnu.org/ml/gcc/2015-11/msg00075.html

This means bugfixing and no new features unless the
new feature has been previously submitted for review.

So, who has outstanding patches needing reviews and
who wants to jump on the squashing of old bug reports?

-- 
Steve

Re: Trunk in Stage 3.

[Paul Richard Thomas]

Dear Steve,

Yes, I did miss it. The deferred character length fix became more than
a little bit obsessive in nature, after I picked it up, and I ignored
nearly everything else that was going on.

I have spent quite some time in the last few months, trying to
engineer a working schema for parameterized derived types. I think
that I have a solution but (i) it will be a fairly monumental
undertaking; and (ii) It will be the sort of addition that is
completely out of line with stage 3. On the bright side, this gives me
a reasonable amount of time to get it right. The submodule work went
this way too. I had been toying with it, when stage 3 of 5 branch
arrived. Although, what I did at that time was entirely wrong headed,
the process of thinking it through meant that implementation went
rather quickly.

Jerry has been working away on user-defined derived-type IO. I had
already undertaken to help him out with this but have fallen short on
that (Sorry, Jerry!). Jerry and I will work together to see if it can
be rolled out on a short timescale.

I have a patch for recursive, allocatable derived-type components more
or less ready to go. I'll see if I can submit it in the next 48 hours.
Of its nature, I believe that it will be save, even for stage 3; ie.
to trigger it requires that such a derived type be declared and the
patch is entirely ring fenced by statements along the lines "if this
component is allocatable and of the same type as the parent derived
type, do such-and-such".

Apart from these two new features, I have loads of bugs to work on!
The oldest concern pointers to components of derived type arrays. I
had hoped to introduce the array descriptor reform but just have not
had the time - it's another rather monumental task! Since it is, to my
knowledge, the only non-compliance with F95, I will submit the kludgy
solution that I have been avoiding.

Cheers

Paul

On 17 November 2015 at 02:13, Steve Kargl
<sgk@troutmask.apl.washington.edu> wrote:
> If you missed the announcement, trunk is now in Stage 3.
>
> https://gcc.gnu.org/ml/gcc/2015-11/msg00075.html
>
> This means bugfixing and no new features unless the
> new feature has been previously submitted for review.
>
> So, who has outstanding patches needing reviews and
> who wants to jump on the squashing of old bug reports?
>
> --
> Steve



-- 
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: Trunk in Stage 3.

[Damian Rouson]

> On Nov 16, 2015, at 7:13 PM, Steve Kargl <sgk@troutmask.apl.washington.edu> wrote:
> 
> If you missed the announcement, trunk is now in Stage 3.
> 
> https://gcc.gnu.org/ml/gcc/2015-11/msg00075.html
> 
> This means bugfixing and no new features unless the
> new feature has been previously submitted for review.
> 
> So, who has outstanding patches needing reviews and
> who wants to jump on the squashing of old bug reports?
> 
> -- 
> Steve



Tobias’s EVENT_TYPE patch is outstanding: https://gcc.gnu.org/ml/fortran/2015-04/msg00119.html.  Alessandro submitted his dissertation draft at the end of October. He expects to have time to look at this patch in the coming weeks and get it submitted for approval for committing to the trunk.

Damian

Re: Trunk in Stage 3.

[Steve Kargl]

Paul,

In truth, I missed the original announcement.  Stumbled acrossed
it today, and thought I should bring it to everyone's attention.

I knew about your PDT effort and Jerry told me on IRC (on Sat.?)
that he was working out the details for DTIO.  Concerning the
array descriptor reform, I recall Tobias indicated that the 
fortran-devel branch had a 90-95% complete implementation.
Unfortunately, its that last 5-10% of effort that is the hard part!

As you may have guessed from my recent uptick in fixing bug reports,
I knew that 6.0 stage 3 was coming.  In addition, Gerhard Steinmetz
has been submitting a number of bugs that sit in my wheel house.  So,
I was simply started to attack the issues I thought I could fix.

Just finished a patch for PR 43966. :-)

Bugzilla (via the gfortran wiki) shows 805 bug reports for Fortran.
I'm hoping a call-to-arms will encourage others to burn any spare
cycles on PRs.  Yeah, I know.  What's a spare cycle as I only closed
1 PR last year! :-(

-- 
steve

On Tue, Nov 17, 2015 at 06:53:25AM +0100, Paul Richard Thomas wrote:
> Dear Steve,
> 
> Yes, I did miss it. The deferred character length fix became more than
> a little bit obsessive in nature, after I picked it up, and I ignored
> nearly everything else that was going on.
> 
> I have spent quite some time in the last few months, trying to
> engineer a working schema for parameterized derived types. I think
> that I have a solution but (i) it will be a fairly monumental
> undertaking; and (ii) It will be the sort of addition that is
> completely out of line with stage 3. On the bright side, this gives me
> a reasonable amount of time to get it right. The submodule work went
> this way too. I had been toying with it, when stage 3 of 5 branch
> arrived. Although, what I did at that time was entirely wrong headed,
> the process of thinking it through meant that implementation went
> rather quickly.
> 
> Jerry has been working away on user-defined derived-type IO. I had
> already undertaken to help him out with this but have fallen short on
> that (Sorry, Jerry!). Jerry and I will work together to see if it can
> be rolled out on a short timescale.
> 
> I have a patch for recursive, allocatable derived-type components more
> or less ready to go. I'll see if I can submit it in the next 48 hours.
> Of its nature, I believe that it will be save, even for stage 3; ie.
> to trigger it requires that such a derived type be declared and the
> patch is entirely ring fenced by statements along the lines "if this
> component is allocatable and of the same type as the parent derived
> type, do such-and-such".
> 
> Apart from these two new features, I have loads of bugs to work on!
> The oldest concern pointers to components of derived type arrays. I
> had hoped to introduce the array descriptor reform but just have not
> had the time - it's another rather monumental task! Since it is, to my
> knowledge, the only non-compliance with F95, I will submit the kludgy
> solution that I have been avoiding.
> 
> Cheers
> 
> Paul
> 
> On 17 November 2015 at 02:13, Steve Kargl
> <sgk@troutmask.apl.washington.edu> wrote:
> > If you missed the announcement, trunk is now in Stage 3.
> >
> > https://gcc.gnu.org/ml/gcc/2015-11/msg00075.html
> >
> > This means bugfixing and no new features unless the
> > new feature has been previously submitted for review.
> >
> > So, who has outstanding patches needing reviews and
> > who wants to jump on the squashing of old bug reports?
> >
> > --
> > Steve
> 
> 
> 
> -- 
> Outside of a dog, a book is a man's best friend. Inside of a dog it's
> too dark to read.
> 
> Groucho Marx

-- 
Steve

Re: Trunk in Stage 3.

[Steve Kargl]

On Tue, Nov 17, 2015 at 12:19:01AM -0600, Damian Rouson wrote:
> 
> 
> > On Nov 16, 2015, at 7:13 PM, Steve Kargl <sgk@troutmask.apl.washington.edu> wrote:
> > 
> > If you missed the announcement, trunk is now in Stage 3.
> > 
> > https://gcc.gnu.org/ml/gcc/2015-11/msg00075.html
> > 
> > This means bugfixing and no new features unless the
> > new feature has been previously submitted for review.
> > 
> > So, who has outstanding patches needing reviews and
> > who wants to jump on the squashing of old bug reports?
> > 
> 
> Tobias???s EVENT_TYPE patch is outstanding:
> https://gcc.gnu.org/ml/fortran/2015-04/msg00119.html
> Alessandro submitted his dissertation draft at the end of October.
> He expects to have time to look at this patch in the coming weeks
> and get it submitted for approval for committing to the trunk.
> 

Thanks for the pointer.  I've lost track of some patches.  I have
a feeling that Andre has one or two unreviewed/uncommitted ALLOCATE
patches.  Tobias's patch is probably about coarrays.  He's unfortunately
our corray expert.  From his absences here, I suspect that Real Life
(tm) has caught up with him.  It's probably a well-deserved respite from
his gfortran hacking.

-- 
Steve