| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | Jan Wieck <JanWieck(at)yahoo(dot)com> |
| Cc: | Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us>, pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: Question about todo item |
| Date: | 2001-08-08 13:45:30 |
| Message-ID: | 1348.997278330@sss.pgh.pa.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Jan Wieck <JanWieck(at)yahoo(dot)com> writes:
> For all the default operations, the system would treat the
> datums still like regular attributes. That means, that an
> INSERT ... SELECT ...
> copying a BLOB from one table to another (and that's correct,
> BLOB's should have copy semantics) would force the entire
> BLOB data into memory ... and ... then ... after ... some
> ... time ... run out of memory.
This does not seem expensive or difficult to solve. tuptoaster.c
will be handed a TOAST pointer as part of heap_insert, and it will
know that it has to duplicate the value. It seems an easy, localized
change to persuade it to do that copying chunk-at-a-time instead of
suck-it-all-in-then-spew-it-all-out.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Jan Wieck | 2001-08-08 13:47:55 | Re: Question about todo item |
| Previous Message | Zeugswetter Andreas SB SD | 2001-08-08 12:46:10 | RE: CURRENT OF cursor without OIDs |