Re: Cleaning up array_in()

13.09.2023 11:55, jian he wrote:
>> 2) CPPFLAGS="-DARRAYDEBUG" ./configure ... breaks "make check", maybe change elog(NOTICE) to elog(DEBUG1)?
>> 2a) a message logged there lacks some delimiter before "lBound info":
>> NOTICE:  array_in- ndim 1 ( 3 -1 -1 -1 -1 -1lBound info 1 1 1 1 1 1) for {red,green,blue}
>> what about changing the format to "ndim 1 ( 3 -1 -1 -1 -1 -1; lBound info: 1 1 1 1 1 1)"?
> fixed. Use elog(DEBUG1) now.

Thanks for the fixes!

I didn't mean to remove the prefix "array_in-", but in fact I was confused
by the "{function_name}-" syntax, and now when I've looked at it closely, I
see that that syntax was quite popular ("date_in- ", "single_decode- ", ...)
back in 1997 (see 9d8ae7977). But nowadays it is out of fashion, with most
of such debugging prints were gone with 7a877dfd2 and the next-to-last one
with 50861cd68. Moreover, as the latter commit shows, such debugging output
can be eliminated completely without remorse. (And I couldn't find mentions
of ARRAYDEBUG in pgsql-bugs, pgsql-hackers archives, so probably no one used
that debugging facility since it's introduction.)
As of now, the output still weird (I mean the excessive right parenthesis):
DEBUG:  ndim 1 ( 2 -1 -1 -1 -1 -1); lBound info: 1 1 1 1 1 1) for {0,0}

Otherwise, from a user perspective, the patch set looks good to me. (Though
maybe English language editorialization still needed before committing it.)

Best regards,
Alexander