11:35:11scymtymstassats: thanks (and sorry for messing up the filesystem condition stuff)
11:37:21stassatsansi tests caught some, you can run ansi-tests with tests/ansi-tests.sh
11:37:31stassatsit'll show the difference from the baseline
11:38:36scymtymi have seen the hash-table problem a few times since upgrading to 1.5.6. all instances were in the context of clx's DISPLAY-RESOURCE-ID-MAP which is a largish (1658 entries in this case) non-synchronized, non-weak EQL hash-table
11:39:26scymtymi still have the session open if anyone has debugging ideas
12:05:56scymtymonly under a lock, as far as i can tell
12:07:19scymtymi can't rule out misuse in clx that is only now caught by additional AVERs
12:07:52stassatsunlikely it could've survived concurrent modification in the past
12:09:38scymtymmaybe the corruption didn't matter for the way clx uses the hash-table. this is also very rare, only happened a couple of times since upgrading while using clx all day every day