Re: [patch, fortran] PR43832 OPEN statement not diagnosing missing unit number

[Jerry DeLisle <jvdelisle at verizon dot net>]

Hans-Peter Nilsson wrote:
> On Fri, 23 Apr 2010, Jerry DeLisle wrote:
> > Jerry DeLisle wrote:
> > > Hi folks,
> > >
> > > The attached patch adds an error message. I also found a redundant and bad
> > > test case fgetc_3.f90 that failed with this patch and so I deleted it and
> > > added the new test case open_nounit.f90.
> > Disregard the fgetc_3.f90 missing. My local trunk was at r158397.  Otherwise
> > all is identical.
>
> I'm not sufficiently sure (well ok, safe-side thickness) I
> understand you correctly, but in r158687 gfortran.dg/fgetc_3.f90
> is there and failing (for cris-elf, but I guess universally).
> It's not mentioned in your testsuite changelog.
>
> Should it just be removed?
>
> brgds, H-P

To clarify, I found fgetc_3.f90 in my local tree at rev 158397 on my laptop.  I 
discovered it because it failed with the patch.  Looking closer at it I could 
see it had an OPEN statement with a missing UNIT.  Looking at fgetc_2 and 1 I 
could see it was a duplicated test case except for the bad OPEN statement.  So I 
 deleted it.

Then I ported my patch over to my desktop machine from where I do my patch 
commits and found the patch buggered up because fgetc_3.f90 was not on trunk. 
This is when I realized I did not have latest trunk on my laptop and that since 
158397, fgetc_3.f90 had already been deleted from trunk.

This whole fgetc_3.f90 is a bit odd.  I did not find it in my local version of 
158687.

Regardless, it should be deleted.  I do not understand why you find it in your 
trunk.  "svn update" should delete it.  I did not explicitly or implicitly 
delete it from trunk when I committed the open_nounit.f90 test case. "svn 
status" does not show it missing from my local trunk

Hope this clarifies.

Jerry