#75 closed defect (wontfix)
Fix/workaround for the PACK bug in gfortran
Reported by: | kilian | Owned by: | kilian |
---|---|---|---|
Priority: | P1 | Milestone: | |
Component: | core | Version: | 2.0.0alpha |
Severity: | minor | Keywords: | |
Cc: |
Description
The evaluator code now contains references to the PACK routine, which is still buggy in gfortran 4.5. The result is a segfault. Either wait for a gfortran bug fix, or write a replacement routine.
Change History (10)
comment:1 Changed 15 years ago by
Milestone: | v2.0-beta → v2.0-final |
---|---|
Priority: | P2 → P3 |
Severity: | blocker → normal |
comment:2 Changed 15 years ago by
The bugzilla URL for that gfortran bug is http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41478. It is not even assigned right now (should be fixed for the final 4.5.0 release). As yet, I don't see any consequences for 2.0.0_beta!?
comment:3 Changed 15 years ago by
Priority: | P3 → P5 |
---|---|
Severity: | normal → minor |
There is movement again from the gfortran side. I guess, for the future there is nothing to from our side. I leave the ticket open for now as a reminder.
comment:4 Changed 15 years ago by
Tonight this gfortran bug (PR41478) has been assigned to Paul Thomas. It seems that he provided a solution and that this issue can be closed soon. Anyhow, at the moment it doesn't seem to affect the WHIZARD running at all.
comment:5 Changed 15 years ago by
According to the gfortran team, Paul Thomas has fixed this bug two days ago. Right now, this cannot be tested because of bug report #132.
comment:6 Changed 15 years ago by
I've looked into it, seems that now it's working, WK could you double-check, plz.
comment:8 Changed 15 years ago by
Priority: | P5 → P1 |
---|
comment:9 Changed 15 years ago by
Resolution: | → wontfix |
---|---|
Status: | new → closed |
With the gfortran svn version of today, the PACK bug is still present: if I reactivate the critical line, the program crashes with a double-free condition.
However, within WHIZARD the critical code has been replaced by a workaround anyway. We should investigate this and keep contact with the gfortran team, but for the WHIZARD project it is irrelevant. I close the ticket.
This bug seems to have no effect on the WHIZARD tests. (But is not yet resolved in the gfortran trunk ?). This bug should be fixed until the final release of gfortran 4.5.0. So I move this report to the next milestone, or should we close it!?