[Patch, Fortran] PR fortran/35837: Fix type-resolving and gfc_current_ns

[Tobias Burnus]

Hi all,

sorry for writing in German to the list :-(

As service roughly same in English - though there is not much use for
middle-enders and for Fortraners, only the current list of regressions
might be of interest.

Tobias


> Daniel Kraft wrote:
> > Sounds like a good idea to me! And yes, 5 months sound really a bit long...

Indeed, but this does not depend on gfortran as it is not release critical.
 
> > For the final 4.4 release, there'll be another delay afters
> > branching off the 4.4 release, won't there?

Presumably not. For 4.3 stage3 was followed by the release free. And the
release happend immediately after the branching.

(One has learnt from 4.2. Here, after branching everyone jumpt to 4.3
after the gcc-4_2-branch existed. That reduced the time spend on bug fixing
and increased the work as bug fixes had to checked in into the branch and into
the trunk.)

Release criterion is by the way the number of regressions; Fortran is not
release critical, but one may assume that the release will be postponed by
a few days when a serious Fortran regression has been found. (Which has
fortunately not yet happened shortly before a release.)

(Performance regression and compile-time regressions also count as regressions.)


By the way, the current Fortran regressions are relatively unimportant:
  36463 - no true regression
  37274 - has a patch (Paul's)
  36599 - checked in for 4.4 (4.3 not yet)
  35680, 36700 - ice-on-INvalid-code
  36700 - rejects-valid bug? not reproducible
  36599 - Performance bug; Status unclear. Only 4.3?


Stage3 goal for Fortran: No regressions and no Fortran 95 bugs.
(Per definition of Joost [who created the meta bug]: Issues with
standard-conforming Fortran 95 programs, which give wrong results
or are rejected - as long as the issue is platform independent and
occurs with default options.)

Let's see how we will categorize the bugs when no F95 and no regression
will exist. Well, there remain enough bugs to be fixed ;-)

Regarding P1-P3:
 http://gcc.gnu.org/ml/gcc/2008-09/msg00013.html
 P1: 9 PRs, P2: 131 PRs, P3: 28 PRs.

Regards,

Tobias