pg_dump & ownership (again)

Philip Warner <pjw@rhyme.com.au> writes:

Recent testing of the BLOB restoration code of pg_dump has highlighted
something that *may* be a problem, but I am not sure. BLOBs create tables,
and AFAICT those tables are owned by the user who was connected at the
time. This theoretically means that even though a user has access to a
table with a BLOB column, they may not be able to read the BLOB - is that
right? Should I worry about this?

The BLOB access code contains no permissions checks --- which is likely
a bug in itself, but anyway if you know the OID of a large object you
can do anything you want to it via the lo_xxx functions.

regards, tom lane

At 15:49 1/08/00 -0400, Tom Lane wrote:

The BLOB access code contains no permissions checks --- which is likely
a bug in itself, but anyway if you know the OID of a large object you
can do anything you want to it via the lo_xxx functions.

This is good to know; I really hope that a new BLOB implementation can come
out of TOAST, and get rid of this mess ASAP...

----------------------------------------------------------------
Philip Warner | __---_____
Albatross Consulting Pty. Ltd. |----/ - \
(A.C.N. 008 659 498) | /(@) ______---_
Tel: (+61) 0500 83 82 81 | _________ \
Fax: (+61) 0500 83 82 82 | ___________ |
Http://www.rhyme.com.au | / \|
| --________--
PGP key available upon request, | /
and from pgp5.ai.mit.edu:11371 |/