> Hi Tatsuo,
>
> Sorry to bother you but I noticed in the HISTORY file of Postgres-7.0
> this entry relating to 6.5.2:
>
> Fix for large object write-in-middle, no extra block, memory consumption(Tatsuo)
>
> I am using 6.5.2 (7.0 proved too buggy)
What kind of bugs have you found with 7.0?
> but I am getting this problem with
> the Large Object interface:
>
> I lo_lseek to a middle position, lo_read a header of my record,
> lo_read the rest according to size in header, lo_lseek to previous position,
> then lo_write an altered record: any position not written to in the LO
> at the time is trashed and bears no resemblance to previous content.
> As if the Lob FD no longer referred to the right object just garbage.
> I think it is the combination of calls prior to the write, because I
> do single seek/writes elsewhere in my app no problem.
>
>
> I am wondering whether this is the problem you fixed and which version
> the fix actually went into!?
>
> yours sincerely,
> Chris Quinn
Ok, give me some time to dig into problems you mentioned.
(I'm very busy this month).
--
Tatsuo Ishii