Re: [INTERFACES] Re: DELETE FROM TABLE doesn't work (AGAIN) - Mailing list pgsql-hackers

Gerald Fischer wrote:

> Hi David!
>
> On Mon, 07 Sep 1998 10:56:21 -0400, David Hartwig wrote:
>
> First of all I want to thank you for your mail.
>
> >> I reinstalled Postgresql with this patch about 10 times in the last 2 days, but without success. :-(
> >>
> >> I tried now to use the snapshot, and it compiled nearly without problems (in
/src/interfaces/ecpg/preproc/preproc.yis a ';' missing on line 1562), but the ODBC-Driver 
> can't connect to the database. The logfile
> >> says:
> >> ----------------
> >The snapshot is very unstable at this time.    6.4 will not be official until after Oct.1.   I would suggest not
usingit at this time 
>
> Well, I gave it a try.
>
> >> ---------
> >> And with the snapshot of sept. 6th, it complains about a missing MSysConf-Table. :-(
> >> --------
> >> conn=71501948, query='declare SQL_CUR71516440 cursor for SELECT Config, nValue FROM MSysConf'
> >> ERROR from backend during send_query: 'ERROR:  msysconf: Table does not exist.'
> >> --------
> >
> >This is a normal error.  MS Access always queries this table.  It is not necessary for successful processing.
>
> Typically M$ :-(
>
> >>
> >>
> >> Please help, because I would need a running system tomorrow :-( (For the moment it will do it without delete, but
notfor ever). 
> >>
> >
> >It is difficult for me to figure out  what your problem is.   What we need is a reproducible sequence of events
leadingup to the crash. 
>
> Well, no problem :-) I can reproduce the crash. With or without the ODBC-Driver.
>
> >If you can reproduce the crash through the psql monitor, then we should work with that.   No sense in adding an
extralayer of complexity by working through the ODBC 
> driver.
> >
> >The same goes for the junkfilter patch.   The errors as you describe, do not point to the junkfilter patch.
Admittedly,it is possible.   See if the problem can be reproduced 
> without the junkfilter.
>
> Ok, once again.
>
> I've installed postgresql 6.3.2 with all patches from ftp.postgresql.org/pub/patches. It worked great, but I could
notuse the ODBC-Driver for stuff like order by and group by 
> (this is the bug where the junkfilter is a workaround, i think).

Correct.   You may want to make sure you have the latest Jet Engine from M$.     That are serious bugs in older
versions.  See our FAQ at 
http://www.insightdist.com/psqlodbc         This will not solve our current problem.

>
> Therefore I installed the junkfilter, compiled again, and now I was able to order/group over the ODBC-Driver.
> At this time I was happy :-), but than I wanted to make a destroyuser and the backend crashed. I thought it was the
destroyuser-script,but then I tryed to make a simple 
> 'delete from test;' and this crashed the backend, too :-(.
> A few minutes ago we found out that triggers wont work, too.
> Ok, I hoped it was another patch that does not like the junkfilter and I tryed to recompile a completly new version
(onlywith the junkfilter applied). The same error occured. 

Hmm...  Something is not right here.    I have not problem on either AIX or Caldera Linux and I know others are using
thepatch without any problems.    I doubt if I are going to be 
able to resolve this soon.  I am juggling too many things right now.    Our best bet may be to wait until 6.4


>
> Therefore I started today to give the snapshot a try, but it will not work together with the ODBC-Driver :-(.

> Here a capture of psql:
> -----
> test=> \d test
>
> Table    = test
> +----------------------------------+----------------------------------+-------+
> |              Field               |              Type                | Length|
> +----------------------------------+----------------------------------+-------+
> | i                                | int4                             |     4 |
> | j                                | int4                             |     4 |
> +----------------------------------+----------------------------------+-------+
> test=> select * from test;
>   i|  j
> ---+---
> 100|100
> (1 row)
>
> test=> delete from test;
> PQexec() -- Request was sent to backend, but backend closed the channel before r
> esponding.
>         This probably means the backend terminated abnormally before or while pr
> ocessing the request.
> test=>
> test=> select * from test;
> PQexec() -- There is no connection to the backend.
> ------
>
> I hope you can help me, because the only thing I could try is to get another linux-distribution and try it with eg
redhat(at the moment I use SuSE). 
>
>

I am not a Linux guru.  So I don't have a clue a to weather this is an issue.

>
>
> PS: Sorry for my bad english...

It way better that my second language.