Thread: BUG #2962: 8.2.1 lo_creat Documentation incorrect?

The following bug has been logged online:

Bug reference:      2962
Logged by:          Gary Chambers
Email address:      gwchamb@gmail.com
PostgreSQL version: 8.2.1
Operating system:   RHEL WS 4 (2.6.9-42.0.3.EL)
Description:        8.2.1 lo_creat Documentation incorrect?
Details:

8.2.1 documentation, section 30.4:

SELECT lo_creat(-1);       -- returns OID of new, empty large object

The documentation states that a new, empty oid is returned if -1 is passed
to either lo_creat or lo_create.

This returns 2^32-1 on every invocation.  Passing a zero seems to work
correctly, however.

"Gary Chambers" <gwchamb@gmail.com> writes:
> 8.2.1 documentation, section 30.4:
> SELECT lo_creat(-1);       -- returns OID of new, empty large object

That works for me ...

> The documentation states that a new, empty oid is returned if -1 is passed
> to either lo_creat or lo_create.

It says no such thing about lo_create().  Now AFAICS lo_creat() doesn't
pay any attention to its argument --- the fact that it even takes one is
historical.  So -1 or any other input value would behave the same.
But lo_create(-1) would try to create a LO numbered 2^32-1 which would
work the first time and not thereafter.

> This returns 2^32-1 on every invocation.  Passing a zero seems to work
> correctly, however.

Looking at the code, it appears that lo_create(0) will behave like
lo_creat(), that is create a LO with an unspecified OID.

Your report seems a bit confused.  Want to test a little more and
try again?

            regards, tom lane

> Your report seems a bit confused.  Want to test a little more and try again?

Nonsense!  My report is not confused at all.  The reporter, on the
other hand, is -- was!  I apologize for the bogus report.  I hope to
someday redeem myself for my inattention to detail.

-- Gary Chambers

// Nothing fancy and nothing Microsoft!