On Sun, Nov 24, 2013 at 11:09:57AM -0500, Tom Lane wrote:
> Does it? In the example at hand, the questionable comment was after some
> C code and before some SQL code. I'd say it's just about exactly 50-50
Yes, it does. Everything between EXEC SQL and the semicolon is SQL by
definition, the rest is C.
> odds as to whether the user will think that C or SQL conventions should
> apply at that spot. I remain of the opinion that changing behavior here
> will annoy more people than it will help.
I don't agree. Please keep in mind that ecpg can spit out illegal C-code as it
is now. It accepts and outputs nested comments that the C compiler obviously
does not accept. Or in other words, I cannot see that people have code that's
broken by the change I made.
BTW I did notice that the patch is incomplete, though. I need to add another
another check or the nested comments still get put into the C file.
> In any case, you won't make any friends at all unless you can document
> exactly what the behavior is (and I'm disappointed to see that your
> behavior-changing commit did not touch the documentation). There might
> be some reason to change if the new behavior is more easily explainable
> than the old.
I'm absolutely willing to check and update the docs, but didn't so far because
I see this as a bug that got fixed and not a new or changed feature.
So far there is nothing about comments in the docs at all AFAICT. I'd expect
people to expect standard C behaviour.
Michael
--
Michael Meskes
Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org)
Michael at BorussiaFan dot De, Meskes at (Debian|Postgresql) dot Org
Jabber: michael.meskes at gmail dot com
VfL Borussia! Força Barça! Go SF 49ers! Use Debian GNU/Linux, PostgreSQL