From: | Andrew McNamara <andrewm(at)object-craft(dot)com(dot)au> |
---|---|
To: | pgsql-hackers(at)postgresql(dot)org |
Cc: | Andrew McNamara <andrewm(at)object-craft(dot)com(dot)au> |
Subject: | No sanity checking performed on binary TIME parameters. |
Date: | 2009-05-25 04:20:44 |
Message-ID: | E419F08D-B908-446D-9B1E-F3520163CE9C@object-craft.com.au |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
When submitting a query via the V3 binary protocol (PQexecParams,
paramFormats[n]=1), it appears the PostgreSQL server performs no range
checking on the passed values. Passing values greater than 24 hours
results in unpredictable results (dumps that cannot be restored,
strange output when printing the column in psql, etc). Tested with
version 8.1 and 8.2 (integer_datetimes is false).
Using my python ocpgdb module (http://code.google.com/p/ocpgdb/):
>>> db.execute('select %s::time::text', DateTimeDelta(0,23,59,59))
[('23:59:59',)]
>>> db.execute('select %s::time::text', DateTimeDelta(0,28,0,0))
[('K|\x1f',)]
ocpgdb has a lower-level API which is a thin layer on top of libpq -
exercising this directly to rule out any problems with the
mx.DateTime.DateTimeDelta class yields the same results:
>>> import struct
>>> import ocpgdb, oclibpq
>>> db=oclibpq.PgConnection('')
>>> list(db.execute('select $1::time::text', [(ocpgdb.pgoid.time,
struct.pack('!d', 23*60*60))]))
[(<PyPgCell name 'text', type 25, modifier -1, value '23:00:00' at
0x42a4a0>,)]
>>> list(db.execute('select $1::time::text', [(ocpgdb.pgoid.time,
struct.pack('!d', 48*60*60))]))
[(<PyPgCell name 'text', type 25, modifier -1, value 'K|\x1f' at
0x42a500>,)]
Apologies if this bug has already been addressed - I didn't find any
references to it while googling.
From | Date | Subject | |
---|---|---|---|
Next Message | Joshua Tolley | 2009-05-25 04:47:29 | Re: generic options for explain |
Previous Message | Michael Meskes | 2009-05-25 03:24:40 | Warnings in compile |