On Sat, 2005-02-19 at 14:35 -0500, Tom Lane wrote:
Good catch. I've applied the attached patch (this is against 8.0/CVS
tip but applies with some fuzz to 7.4).
Is there a way to repro this via SQL? (It would be nice to have a
regression test...)
-Neil
---(end of
Neil Conway [EMAIL PROTECTED] writes:
On Sat, 2005-02-19 at 14:35 -0500, Tom Lane wrote:
Good catch. I've applied the attached patch (this is against 8.0/CVS
tip but applies with some fuzz to 7.4).
Is there a way to repro this via SQL? (It would be nice to have a
regression test...)
No,
Summary: I can crash 7.4-CVS and 8.0/HEAD by sending what appears to be
a valid query.
I'm trying to insert a unique entry into this table:
create table bodyparts (
id serial primary key,
bytes integer not null,
lines integer not null,
At 2005-02-19 21:38:56 +0530, [EMAIL PROTECTED] wrote:
Summary: I can crash 7.4-CVS and 8.0/HEAD by sending what appears to be
a valid query.
A couple of things I forgot to mention:
1. I turned the logging all the way up and I've uploaded the messages
from the crashing backend to
At 2005-02-19 21:38:56 +0530, [EMAIL PROTECTED] wrote:
I couldn't reproduce this with SQL PREPARE/EXECUTE statements in psql,
so I conjectured that it may have something to do with the parameter
types being unspecified. I added a statement-describe message between
the parse and the bind,
At 2005-02-19 23:18:01 +0530, [EMAIL PROTECTED] wrote:
If I change my code (by means of an unspeakably vile hack, but never
mind that) to specify each parameter types in the parse message, the
server no longer crashes.
In fact, it works if I specify only the two integer types, and leave the
Abhijit Menon-Sen [EMAIL PROTECTED] writes:
Summary: I can crash 7.4-CVS and 8.0/HEAD by sending what appears to be
a valid query.
Good catch. I've applied the attached patch (this is against 8.0/CVS
tip but applies with some fuzz to 7.4).
regards, tom lane
***