PARANOIA_LEVEL
Regina Obe
lr at pcorp.us
Thu Jan 18 17:11:03 PST 2024
Yah lets do that.
From: Paul Ramsey <pramsey at cleverelephant.ca>
Sent: Thursday, January 18, 2024 6:50 PM
To: Sandro Santilli <strk at kbt.io>
Cc: Regina Obe <lr at pcorp.us>; PostGIS Development Discussion <postgis-devel at lists.osgeo.org>
Subject: Re: PARANOIA_LEVEL
I feel like the most “standard” thing to do would be to convert them all to assert()s and then have a
—enable-assert
option. And then we can add other asserts to our heart’s content without having to know about this weird option Dave invented back in 2004.
P
On Jan 18, 2024, at 2:25 PM, Paul Ramsey <pramsey at cleverelephant.ca <mailto:pramsey at cleverelephant.ca> > wrote:
There are only a handful of uses (grep for "PARANOIA_LEVEL >”) and except for one they are all testing > 0. So I think an on/off, or potentially just strip them out. I never flip them on, on purpose, and have never assumed anything other than that it was always off (imagine my surprise to find it on).
P
On Jan 18, 2024, at 2:23 PM, Sandro Santilli <strk at kbt.io <mailto:strk at kbt.io> > wrote:
On Thu, Jan 18, 2024 at 04:25:55PM -0500, Regina Obe wrote:
On Thu, Jan 18, 2024 at 08:13:04AM -0800, Paul Ramsey wrote:
Just a reminder, if you build postgis with —enable-debug, the
PARANOIA_LEVEL gets bumped up, and that can have huge performance effects
which make any kind of benchmarking a whack-a-mole process.
I guess I internalized the idea that —enable-debug would just drop a ‘-g’ flag
into the build and that’s about it, but in fact it has these other deleterious effects.
Among other things, it makes lwcollection_add_lwgeom a couple orders of
magnitude slower.
Anyways, super important for anyone building for release too. You aren’t just
getting debug symbols!
I wonder if we should split off —enable-debug from other “developer
affordances”, in a different option, kind of like how pgsql as —enable-cassert and
some other #define only options for things like memory checking and so on.
+1 --enable-paranoia would sound good
+1
Actually I guess it should be --with-paranoia-level N
--strk;
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-devel/attachments/20240118/477d2d7b/attachment.htm>
More information about the postgis-devel
mailing list