Peter Eisentraut wrote:
> There is also one need error that needs further investigation.
I've looked at this bit in the regression tests about \gexec:
--- a/src/test/regress/expected/psql.out
+++ b/src/test/regress/expected/psql.out
@@ -232,11 +232,7 @@ union allselect 'drop table gexec_test', 'select ''2000-01-01''::date as party_over'\gexecselect 1
asones
- ones
-------
- 1
-(1 row)
-
+ERROR: DECLARE CURSOR can only be used in transaction blocks
This can be interpreted as two separate errors:
* \gexec ignores the first result
postgres=# select 'select 1','select 2' \gexec?column?
----------2
(1 row)
* \gexec fails with FETCH_COUNT
postgres=# \set FETCH_COUNT 1 postgres=# select 'select 1','select 2' \gexec ERROR: DECLARE CURSOR can only be used
intransaction blocks ?column? ---------- 2 (1 row)
The two issues are due to SendQuery() being reentered
for the gexec'd queries when it hasn't finished yet with the
main query.
I believe that just collecting all results of \gexec before
executing any of them would solve both errors.
Also doing a bit more testing I've seen these other issues:
* combining multiple result sets and FETCH_COUNT doesn't work:
postgres=# \set FETCH_COUNT 1 postgres=# select 1 \; select 2; postgres=#
* last error is not recorded for \errverbose :
postgres=# select foo; ERROR: column "foo" does not exist LINE 1: select foo; ^ postgres=# \errverbose There is
noprevious error.
* memory leaks on PGResults.
Best regards,
--
Daniel Vérité
PostgreSQL-powered mailer: http://www.manitou-mail.org
Twitter: @DanielVerite