Re: Is GFortran stable enough to actually use?

["Christopher D. Rickett" <crickett at lanl dot gov>]

On Mon, 18 Sep 2006, Steve Kargl wrote:

> On Mon, Sep 18, 2006 at 02:23:15PM -0600, Christopher D. Rickett wrote:
> > i did an svn update today and fixed the bug of mine that was causing 
> > char_result_11.f90 to fail, but now i'm seeing the following:
> > 
> > FAIL: gfortran.dg/spread_shape_1.f90 (test for excess errors)
> > 
> > this is the only unexpected failure i'm seeing.  this wasn't happening on 
> > friday, so i take it this is new?
> > 
> > Chris
> > 
> 
> It's a new test that pault committed.  Looks like he made
> a follow up commit to fix a problem with the test.  Try
> updating testsuite/, gcc/fortran, and libgfortran/ again.
> You should be able to build without a bootstrap.
> 
> With your fix for char_result_11.f90, does this mean we are
> closer to seeing an ISO_C_BINDING patch?
> 

i'll try updating again and retesting.  the bug of mine breaking 
char_result_11.f90 was pretty simple; i have a flag for the symbols i 
generate that are defined by iso_c_binding, and i was not clearing it.  
this caused a junk value to possibly get written into the .mod file.  i 
also made a slight change to the tree node for by-value params; the 
by-value stuff was working before, but a tree node was still being created 
that hid the true type behind a reference type.  this was allocating 
unecessary space for by-value params.  i've now prevented the reference 
nodes from being allocated.  my value tests still work as before, so i 
think the change is fine.

the code is pretty close, i think.  i'm not making any significant 
changes or additions, since it's too close for me to check that it 
works if i did change something.  i'm still trying to create dejagnu 
tests for some of the test cases i have, but converting some of them is a 
little difficult because the tests are for verifying that gfortran catches 
certain errors.  also, i'm trying to get ahold of the FSF person that was 
working with me on my paperwork, but the individual is out of the office 
today.  i'll call them again tomorrow.  :-)

Thanks.
Chris