| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | Sean Chittenden <sean(at)chittenden(dot)org> |
| Cc: | pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: AutoCommit GUC breaks CLI tools... |
| Date: | 2002-09-02 05:36:18 |
| Message-ID: | 6871.1030944978@sss.pgh.pa.us |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Sean Chittenden <sean(at)chittenden(dot)org> writes:
> This is pretty chump and easy to get around, but it took me a sec to
> figure this out. Anyway, the short and skinny being that with the new
> AutoCommit GUC turned off, create(lang|db) won't work until you bail
> out of the transaction. A quick hack would be to insert an "ABORT;"
> in each of the CLI tools. Anyone have any thoughts on the best way to
> solve this? -sc
Well, like I was saying earlier, I think there is a *lot* of client-side
code that is not ready for this. I would not have thrown in the
autocommit backend feature at all, except that we need it to run the
NIST SQL-spec-compliance tests.
The answer for createlang is probably to throw a BEGIN and a COMMIT
around each command --- that should make it work in either autocommit
on or off modes. We'll have to see how well that scales to other stuff.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Bruce Momjian | 2002-09-02 05:37:23 | Re: AutoCommit GUC breaks CLI tools... |
| Previous Message | Bruce Momjian | 2002-09-02 05:20:59 | Re: RULE regression test failure |