Re: GUC values - recommended way to declare the C variables? - Mailing list pgsql-hackers

From Michael Paquier
Subject Re: GUC values - recommended way to declare the C variables?
Date
Msg-id Y1tC7RsFfSpP4/4M@paquier.xyz
Whole thread Raw
In response to Re: GUC values - recommended way to declare the C variables?  (Peter Smith <smithpb2250@gmail.com>)
Responses Re: GUC values - recommended way to declare the C variables?  (Michael Paquier <michael@paquier.xyz>)
Re: GUC values - recommended way to declare the C variables?  (Michael Paquier <michael@paquier.xyz>)
List pgsql-hackers
On Thu, Oct 27, 2022 at 07:00:26PM +1100, Peter Smith wrote:
> The GUC defaults of guc_tables.c, and the modified GUC C var
> declarations now share the same common #define'd value (instead of
> cut/paste preprocessor code).

Thanks.  I have not looked at the checkup logic yet, but the central
declarations seem rather sane, and I have a few comments about the
latter.

+#ifdef WIN32
+#define DEFAULT_UPDATE_PROCESS_TITLE false
+#else
+#define DEFAULT_UPDATE_PROCESS_TITLE true
+#endif
This is the kind of things I would document as a comment, say
"Disabled on Windows as the performance overhead can be significant".

Actually, pg_iovec.h uses WIN32 without any previous header declared,
but win32.h tells a different story as of ed9b3606, where we would
define WIN32 if it does not exist yet.  That may impact the default
depending on the environment used?  I am wondering whether the top of
win32.h could be removed, these days..

+#ifdef USE_PREFETCH
+#define DEFAULT_EFFECTIVE_IO_CONCURRENCY 1
+#define DEFAULT_MAINTENANCE_IO_CONCURRENCY 10
+#else
+#define DEFAULT_EFFECTIVE_IO_CONCURRENCY 0
+#define DEFAULT_MAINTENANCE_IO_CONCURRENCY 0
+#endif
These don't make sense without prefetching available.  Perhaps that's
obvious enough when reading the code still I would add a small note.
--
Michael

Attachment

pgsql-hackers by date:

Previous
From: Thomas Munro
Date:
Subject: Merging LatchWaitSet and FeBeWaitSet
Next
From: Michael Paquier
Date:
Subject: Re: GUC values - recommended way to declare the C variables?