Horton

Re: TupleTableSlot abstraction

Andres Freund <andres@anarazel.de>

View thread View on postgr.es
From: Andres Freund <andres@anarazel.de>
To: Michael Paquier <michael@paquier.xyz>
Cc: Jeff Janes <jeff.janes@gmail.com>, pgsql-hackers <pgsql-hackers@postgresql.org>, Robert Haas <robertmhaas@gmail.com>, Haribabu Kommi <kommi.haribabu@gmail.com>, Amit Khandekar <amitdkhan.pg@gmail.com>, Alvaro Herrera <alvherre@2ndquadrant.com>, Amit Kapila <amit.kapila16@gmail.com>
Date: 2019-02-27T05:42:38Z
Lists: pgsql-hackers 
Hi,

On 2019-02-27 14:21:52 +0900, Michael Paquier wrote:
> On Sat, Feb 16, 2019 at 05:07:44PM -0500, Jeff Janes wrote:
> > By blind analogy to the changes made to matview.c, I think that createas.c
> > is missing a heap_freetuple, as attached.

First, sorry to have been slow answering. I was whacking around code
further modifying this, and thought I'd just solve the immediate issue
here by committing the followup work that removes getting the tuple out
of the slot entirely. That took longer than planned, so it makes sense
to commit an interim fix.


> I think that you are right.  CTAS and materialized views share a lot
> when it comes to relation creation and initial table loading.  I have
> reproduced the leak and could notice that your fix is correct.  So
> committed.

I'm not so sure that's the architecturally correct fix however. Is it
actually guaranteed, given expanded tuples, toasting, etc, that there's
no other memory leak here? I wonder if we shouldn't work twoards using a
short lived memory context here. Note how e.g. printtup() uses a short
lived context for its work.

Greetings,

Andres Freund

Commits

  1. Fix memory leak when inserting tuple at relation creation for CTAS

  2. Build HashState's hashkeys expression with the correct parent.

  3. Move TupleTableSlots boolean member into one flag variable.

  4. Move generic slot support functions from heaptuple.c into execTuples.c.

  5. Remove absolete function TupleDescGetSlot().

  6. Remove function list from prologue of execTuples.c.

  7. Split ExecStoreTuple into ExecStoreHeapTuple and ExecStoreBufferHeapTuple.

  8. Change TupleTableSlot->tts_nvalid to type AttrNumber.

  9. Deduplicate code between slot_getallattrs() and slot_getsomeattrs().

  10. Mention ownership requirements for REFRESH MATERIALIZED VIEW in docs

  11. Spell "partitionwise" consistently.

  12. Fix WHERE CURRENT OF when the referenced cursor uses an index-only scan.