[Fwd: GCC 4.4.0 Status Report (2008-08-22)]

[Tobias Burnus]

Stage3 is approaching fast (10 days?). In my opinion, we should try to 
get in at least:
- The Finalize patch
- Type-bound procedure work (even if incomplete*)
- Procedure components (even if incomplete*)

(* incomplete implies still usable and giving proper error messages if 
one hits unimplemented stuff -- parser only is also OK; for corner 
cases, having only a PR could be enough.)

Note: The rules allow that patches which have been posted before Stage3 
has started still may be checked in in the early days of Stage1 (incl. 
review fixes).

  * * *

In stage 3, we should try to do the same as for 4.3 - to get the number 
of Fortran 95 bugs down to zero, the number of regressions to zero and 
fix some other bugs. Joost has recently updated the Fortran 95 bug list 
(currently 24 PRs):
http://gcc.gnu.org/bugzilla/showdependencytree.cgi?id=32834&hide_resolved=1

Regression count is the same as last time:
a) 36599 major execution regression for induct.f90 polyhedron benc...
b) 36371 [4.3 Regression] Wrong locus for errors in DATA statement
c) 36700 [4.3/4.4 Regression] ICE on calling a function
d) 35680 [4.3/4.4 regression] ICE on invalid transfer in variable ...
e) 36463 [4.4 Regression] ICE in expand_expr_real_1, at expr.c:726...

In total nothing to worry about: one ICE-on-valid for programs which 
were uncompilable before; two ice-on-invalid, one 4.3/diagnostic-only 
with patch, and one performance regression, which might be rather a 
middle-end problem.

Tobias
-------------- next part --------------
An embedded message was scrubbed...
From: Richard Guenther <rguenther@suse.de>
Subject: GCC 4.4.0 Status Report (2008-08-22)
Date: Fri, 22 Aug 2008 17:46:04 +0200 (CEST)
Size: 3377
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20080822/3c408f66/attachment.eml>