hyc@symas.com wrote:

DG> We've scripted a test which works for me against 2.4.8 and did work DG> against CVS HEAD until Howard committed

Apologies for the vagueness of last night's email.

For a script whose aim is to demo a crash, "works" means "reproduces the crash with no more intervention than invoking it and pressing enter when prompted". I was trying to explain that in cases where replication doesn't occur, the test can't be done, but on re-reading in daylight perhaps I failed on that score ;-)

There were two commits five minutes apart. I could reproduce the crash with syncrepl.c 1.387; replication 2-->1 did not occur with 1.389; I was just about to test with 1.388 when I read your second post.

HYC> syncrepl.c is now updated in HEAD to allow this replication to occur.

HYC> Using your test, I don't see any crash.

Thank you. We'll re-test with current HEAD and post the outcome.

Cheers

Duncan