Re: [PATCH] fix GIN index search sometimes losing results

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Pavel Borisov <pashkin(dot)elfe(at)gmail(dot)com>
Cc: pgsql-hackers(at)lists(dot)postgresql(dot)org, Artur Zakirov <zaartur(at)gmail(dot)com>
Subject: Re: [PATCH] fix GIN index search sometimes losing results
Date: 2020-07-22 15:10:10
Message-ID: 8160.1595430610@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Pavel Borisov <pashkin(dot)elfe(at)gmail(dot)com> writes:
> For 0002-remove-calc-not-flag.patch
> The patch changes the behavior which is now considered default. This is true in RUM module and maybe in some other tsearch side modules. Applying the patch can make code more beautiful but possibly will not give some performance gain and bug is anyway fixed by patch 0001.

I'd be willing to compromise on just adding TS_EXEC_CALC_NOT to the
calls that are missing it today. But I don't see why that's really
a great idea --- it still leaves a risk-of-omission hazard for future
callers. Calculating NOTs correctly really ought to be the default
behavior.

What do you think of replacing TS_EXEC_CALC_NOT with a different
flag having the opposite sense, maybe called TS_EXEC_SKIP_NOT?
If anyone really does need that behavior, they could still get it,
but they'd have to be explicit.

> Overall I'd recommend patch 0001-make-callbacks-ternary.patch and close the issue.

The other issue we have to agree on is whether we want to sneak this
fix into v13, or wait another year for it. I feel like it's pretty
late to be making potentially API-breaking changes, but on the other
hand this is undoubtedly a bug fix.

regards, tom lane

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Alvaro Herrera 2020-07-22 15:36:05 Re: expose parallel leader in CSV and log_line_prefix
Previous Message vignesh C 2020-07-22 14:26:25 Re: Parallel copy