Thread: Re: [INTERFACES] Re: [HACKERS] Access & Postgres
> http://www.insightdist.com/psqlodbc/junkfilter_patch.txt > I know the code works. Hi David. Do you have access to the latest cvs development tree for Postgres? I'm seeing backend crashes from the "junkfilter" regression test, and have been for some time. I (just this morning) applied some patches to the tree to fix up type coersion for inserts from other tables or columns, and this changed code in the "junkfilter" area. However, the regression test behavior is still bad. If you have access to the latest tree, and had time to look at this it would be great. Otherwise, I'll start poking at it. Talk to you soon... - Tom
> I'm seeing backend crashes from the "junkfilter" regression test
I think I have a patch for the problem:
postgres=> select c, count(*) from test_missing_target group by 3;
ERROR: ORDER/GROUP BY position 3 is not in target list
Previously this query provoked a core dump. Will do some regression
testing and then commit to the source tree...
- Tom
Tom, Yes, I do have access to the cvs tree from home. And, I will check it out when I get home. I am curious though. Did you apply the patch, mentioned below, to your latest 6.4 tree? This patch should have been already part of the latest 6.4. And, to my knowledge, it was already applied completely and successfully. Anyway, the intent of the patch on our web site was to allow 6.3 users to be able to use MS Access cleanly. Now that I think of it, though, this version of the patch does not include the new "output" and "sql" files necessary for the regression test. In any case, I should clean that up. Thomas G. Lockhart wrote: > > http://www.insightdist.com/psqlodbc/junkfilter_patch.txt > > I know the code works. > > Hi David. Do you have access to the latest cvs development tree for > Postgres? I'm seeing backend crashes from the "junkfilter" regression > test, and have been for some time. I (just this morning) applied some > patches to the tree to fix up type coersion for inserts from other > tables or columns, and this changed code in the "junkfilter" area. > However, the regression test behavior is still bad. > > If you have access to the latest tree, and had time to look at this it > would be great. Otherwise, I'll start poking at it. Talk to you soon... > > - Tom
> I am curious though. Did you apply the patch, mentioned below, to
> your latest 6.4 tree? This patch should have been already part of
> the latest 6.4. And, to my knowledge, it was already applied
> completely and successfully.
I did my final tests using a fresh cvs tree two days ago just before
committing my big "patch wad". In the last two or three weeks I did have
a "merge problem" with my patches in the resjunk area, which might have
been the patches you are talking about.
> Now that I think of it, though, this version of the patch does not
> include the new "output" and "sql" files necessary for the regression
> test. In any case, I should clean that up.
I am in the process of renaming and moving the "resjunk" test to
"select_implicit". Will try to commit the changes this morning so it is
available to you.
At the moment, all regression tests (except "random") pass on my Linux
box :)
- Tom
> > http://www.insightdist.com/psqlodbc/junkfilter_patch.txt > > I know the code works. > > Hi David. Do you have access to the latest cvs development tree for > Postgres? I'm seeing backend crashes from the "junkfilter" regression > test, and have been for some time. I (just this morning) applied some > patches to the tree to fix up type coersion for inserts from other > tables or columns, and this changed code in the "junkfilter" area. > However, the regression test behavior is still bad. > > If you have access to the latest tree, and had time to look at this it > would be great. Otherwise, I'll start poking at it. Talk to you soon... I have now made the junk filter only run when needed. Perhaps this will fix the problem. -- Bruce Momjian | 830 Blythe Avenue maillist@candle.pha.pa.us | Drexel Hill, Pennsylvania 19026 + If your life is a hard drive, | (610) 353-9879(w) + Christ can be your backup. | (610) 853-3000(h)