armtuk

So stracing postgres shows that my function deg2rad was segfaulting. Wow - apparently some query is calling deg2rad and passing in a null value. The function doesn't like that, so I've put in a test to make sure if it gets a null, it returns a null. So far it's looking promising. I could now submit the admin page that I was using before to test the problem, so my reproducable case has gone away. The other application that was doing it lots was trac. I'm not sure how trac's crashes can be attributed to my stored procedure in an unrelated database, but a bit of clicking around in trac has shown no problems yet. Tomorrow will be the big test. It seems to happen during the day when the server is under load much more prevalently.