public inbox for [email protected]  
help / color / mirror / Atom feed
From: Euler Taveira <[email protected]>
To: Michael Paquier <[email protected]>
To: Dilip Kumar <[email protected]>
Cc: Amit Kapila <[email protected]>
Cc: Alvaro Herrera <[email protected]>
Cc: Petr Jelinek <[email protected]>
Cc: Peter Eisentraut <[email protected]>
Cc: Tomas Vondra <[email protected]>
Cc: Andres Freund <[email protected]>
Cc: Kuntal Ghosh <[email protected]>
Cc: [email protected] <[email protected]>
Cc: pgsql-hackers <[email protected]>
Subject: Re: [BUG]Update Toast data failure in logical replication
Date: Mon, 24 Jan 2022 15:55:34 -0300
Message-ID: <[email protected]> (raw)
In-Reply-To: <[email protected]>
References: <CAA4eK1+k2TKPmdeZxgp2vgFtv2C8+dMKHR=-ty9HswDcCqkK6A@mail.gmail.com>
	<[email protected]>
	<CAA4eK1KgZr=QSBE_Qh0Qfb2ma1Tc6+ZxkMaUHO7aC7b9WSCRaw@mail.gmail.com>
	<CAFiTN-uW50w0tWoUBg_VYCdvNeCzT=t=JzhmiFd452FrLOwMMQ@mail.gmail.com>
	<[email protected]>

On Mon, Jan 24, 2022, at 12:58 AM, Michael Paquier wrote:
> FWIW, I find the API changes of HeapDetermineModifiedColumns() and
> ExtractReplicaIdentity() a bit grotty.  Shouldn't we try to flatten
> the old tuple instead?  There are things like
> toast_flatten_tuple_to_datum() for this purpose if a tuple satisfies
> HeapTupleHasExternal(), or just heap_copy_tuple_as_datum().
> 
I checked v4 and I don't like the HeapDetermineModifiedColumns() and
heap_tuple_attr_equals() changes either. It seems it is hijacking these
functions to something else. I would suggest to change the signature to

static void
heap_tuple_attr_equals(TupleDesc tupdesc, int attrnum,
                       HeapTuple tup1, HeapTuple tup2,
                       bool *is_equal, bool *key_has_external);

and

static void
HeapDetermineModifiedColumns(Relation relation, Bitmapset *interesting_cols,
                             HeapTuple oldtup, HeapTuple newtup,
                             Bitmapset *modified_attrs, bool *key_has_external);

I didn't figure out a cheap way to check if the key has external value other
than slightly modifying the HeapDetermineModifiedColumns() function and its
subroutine heap_tuple_attr_equals(). As Alvaro said I don't think adding
HeapTupleHasExternal() (as in v3) is a good idea because it does not optimize
genuine cases such as a table whose PK is an integer and contains a single
TOAST column.

Another suggestion is to keep key_changed and add another attribute
(key_has_external) to ExtractReplicaIdentity(). If we need key_changed in the
future we'll have to decompose it again. You also encapsulates that
optimization into the function that helps with future improvements/fixes.

static HeapTuple
ExtractReplicaIdentity(Relation relation, HeapTuple tp, bool key_changed,
                       bool key_has_external, bool *copy);


--
Euler Taveira
EDB   https://www.enterprisedb.com/


view thread (5+ messages)  latest in thread

reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Reply to all the recipients using the --to and --cc options:
  reply via email

  To: [email protected]
  Cc: [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected]
  Subject: Re: [BUG]Update Toast data failure in logical replication
  In-Reply-To: <[email protected]>

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox