Re: Push down time-related SQLValue functions to foreign server

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Alexander Pyhalov <a(dot)pyhalov(at)postgrespro(dot)ru>
Cc: Ranier Vilela <ranier(dot)vf(at)gmail(dot)com>, Zhihong Yu <zyu(at)yugabyte(dot)com>, Pgsql Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Push down time-related SQLValue functions to foreign server
Date: 2022-01-21 19:57:12
Message-ID: 1208694.1642795032@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Alexander Pyhalov <a(dot)pyhalov(at)postgrespro(dot)ru> writes:
> So far I have the following prototype. It seems to be working, but I
> think it can be enhanced.
> At least, some sort of caching seems to be necessary for
> is_stable_expr().

Yeah, from a performance standpoint this seems pretty horrid ---
it's probably exponential in the size of the expressions considered
because of the repeated recursions.

The approach I had in mind was to extend the responsibility of
foreign_expr_walker to make it deduce the classification of
an expression in a bottom-up fashion. Also, as I noted before,
we don't want to re-deduce that in a later deparse pass, so
maybe we should just go ahead and deparse during foreign_expr_walker.
Not sure about that part. It sounds expensive; but recording the
classification results in a way we could re-use later doesn't seem
too cheap either.

BTW, the patch is just about unreadable as-is. It would have been
better to not have re-indented the bulk of foreign_expr_walker,
leaving that for some later pgindent pass. But that may be moot,
because I don't think we can tolerate the double-recursion approach
you took here.

regards, tom lane

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Andres Freund 2022-01-21 20:00:57 pg_basebackup fsyncs some files despite --no-sync (was: Adding CI to our tree)
Previous Message Andres Freund 2022-01-21 19:49:56 Re: Avoid erroring out when unable to remove or parse logical rewrite files to save checkpoint work