Horton

Re: tableam vs. TOAST

Peter Eisentraut <peter.eisentraut@2ndquadrant.com>

View thread View on postgr.es
From: Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
To: Robert Haas <robertmhaas@gmail.com>, Andres Freund <andres@anarazel.de>
Cc: Alvaro Herrera <alvherre@2ndquadrant.com>, Prabhat Sahu <prabhat.sahu@enterprisedb.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2019-11-21T10:37:01Z
Lists: pgsql-hackers 
On 2019-11-08 17:59, Robert Haas wrote:
> OK. Could you see what you think of the attached patches? 0001 does
> some refactoring of toast_fetch_datum() and toast_fetch_datum_slice()
> to make them look more like each other and clean up a bunch of stuff
> that I thought was annoying, and 0002 then pulls out the common logic
> into a heap-specific function. If you like this direction, we could
> then push the heap-specific function below tableam, but I haven't done
> that yet.

Partial review: The 0001 patch seems very sensible.  Some minor comments 
on that:

Perhaps rename the residx variable (in both functions).  You have gotten 
rid of all the res* variables except that one.  That name as it is right 
now isn't very helpful at all.

You have collapsed the error messages for "chunk %d of %d" and "final 
chunk %d" and replaced it with just "chunk %d".  I think it might be 
better to keep the "chunk %d of %d" wording, for more context, or was 
there a reason why you wanted to remove the total count from the message?

I believe this assertion

+   Assert(endchunk <= totalchunks);

should be < (strictly less).

In the commit message you state that this assertion replaces a run-time 
check, but I couldn't quite make out which one you are referring to 
because all the existing run-time checks are kept, with slightly 
refactored conditions.

-- 
Peter Eisentraut              http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Commits

  1. tableam: New callback relation_fetch_toast_slice.

  2. tableam: Allow choice of toast AM.

  3. Move heap-specific detoasting logic into a separate function.

  4. Code cleanup for toast_fetch_datum and toast_fetch_datum_slice.

  5. Rename some toasting functions based on whether they are heap-specific.

  6. Create an API for inserting and deleting rows in TOAST tables.

  7. Split tuptoaster.c into three separate files.

  8. tableam: Move heap-specific logic from needs_toast_table below tableam.