Sun 31 Oct 2010 11:15:57 PM UTC, comment #8:
>Not having pending transformations seems a shame. But as they don't seem to currently work properly anyway, this patch is technically an improvement.
I can read some frustration in that comment, and I sympathize with it. I would be happy to, later, restore pending transformations after fixing the underlying problem, and if you want to file a separate bug report to that effect then that would make sense too.
Nevertheless, as you say, this is an improvement, so I pushed it.
|
Sun 31 Oct 2010 12:49:48 PM UTC, comment #7:
Not having pending transformations seems a shame. But as they don't seem to currently work properly anyway, this patch is technically an improvement.
|
Sat 30 Oct 2010 05:33:15 PM UTC, comment #6:
The following patch fixes the crash and the glib criticals, but it also eliminates pending transformations from the GUI. That might not be a bad thing, since they do tend to confuse new users at least.
John, what do you think?
|
Sat 30 Oct 2010 08:59:13 AM UTC, comment #5:
The general problem seems to be that the datastore is left in an inconsistent state when syntax terminates with outstanding transformations.
|
Sat 30 Oct 2010 03:42:48 AM UTC, comment #4:
Actually, running with G_DEBUG=fatal_criticals, it is sufficient to run the original syntax file only once in the GUI to make PSPPIRE die with:
with backtrace:
|
Sat 30 Oct 2010 03:23:05 AM UTC, comment #3:
Applying the following patch causes an assertion failure:
|
Sat 30 Oct 2010 03:19:33 AM UTC, comment #2:
valgrind shows the following
|
Wed 27 Oct 2010 04:37:49 PM UTC, comment #1:
Confirmed here. Also, doesn't crash in the terminal UI whether run once or twice.
|
Sun 08 Aug 2010 05:40:56 PM UTC, original submission:
The following syntax (which contains a error) ...
new file.
data list list /fred * .
begin data
99
99
97
end data.
compute x = 4.
DESCRIPTIVES /rxed .
... crashes on the second attempt to run it in the GUI.
The "compute x = 4." line seems to be relevant.
|