Closed robe2 closed 11 years ago
We do have PgSQL tests, they are in the pgsql directory, 'make installcheck'. however, they are PgSQL-style tests, not PostGIS tests, so they are inferior in all sorts of ways
P
Paul Ramsey http://cleverelephant.ca/ http://postgis.net/
On Tuesday, 2 July, 2013 at 9:57 PM, Regina Obe wrote:
I don't seem to be able to mark this as an enhancement. Anyrate it would be REALLY NICE if we had PostgreSQL tests. I'd even be wiling to help out. Not a rush since I know its harder to do than cunit and you are still in alpha mode. I suspect for example with the Debian issue I am having that the extension won't even install though it passes cunit with flying colors. This would also catch issues with gcc/vc++ differences where the postgresql installed was built with a different gcc than what the user compiled pointcloud with and when the two get combined in PostgreSQL they crash.
— Reply to this email directly or view it on GitHub (https://github.com/pramsey/pointcloud/issues/7).
ah okay great. I'll play around wiith those next to see if I can make them less brittle. Once cunit issues are out of the way.
Are you sure you have this sorta working like on your box or is this an IN THEORY its sorta working on somebody's box? I get a:
make: *** No rule to make target
installcheck'. Stop.`
I haven't had issue make installchecking plv8 and most of Andrew Dunstan's extensions like my favorite smallish unnest_ordinality https://github.com/pgexperts/unnest_ordinality/blob/master/Makefile
I can follow how the regress tests flow into those, but not sure how setting REGRESS=pointcloud on yours could find said tests.
ah nevermind user error. Figured it out you have to cd into the folder.
Looks like you have ght and other tests mixed together so installcheck fails with all sorts of ght not installed. I think we maybe need a separate file for testing ght and conditionally add to the filter of tests.
I'll close this out since you technically have tests and open another more specific ticket.
I don't seem to be able to mark this as an enhancement. Anyrate it would be REALLY NICE if we had PostgreSQL tests. I'd even be wiling to help out. Not a rush since I know its harder to do than cunit and you are still in alpha mode.
I suspect for example with the Debian issue I am having that the extension won't even install though it passes cunit with flying colors. This would also catch issues with gcc/vc++ differences where the postgresql installed was built with a different gcc than what the user compiled pointcloud with and when the two get combined in PostgreSQL they crash.