"Bad VM cache data" message behavior takes hours

[Mark Diekhans]

I was wanted to experiment with the trunk VM. When visiting a mailbox created with 8.2.0b, I get one

:Bad VM cache data: ...": per message.

This is simple because the vector needs to be longer

This is very alarming to see and one doesn't realizes it's not corruption until reading the code. It's at
the most severe warn level and then it waits 2 second per message. I have mail boxes that would take hours
to convert with the 2 second wait.

If the mailbox is read-only, this happens every time the box is visited.

Suggest that the text of the message be made less alarming and that the level be change do it doesn't
show up by default.

[Uday Reddy (reddyuday)]

Apparenty, this also happens when upgrading from 8.1.2 to 8.2.0b. It needs to be painless.

[Uday Reddy (reddyuday)]

This is apparently more serious than it first appeared.

- Looking at a VM 8.2.1 file, such as test-mime, is not only giving tons of Bad VM cache messages, but it is also crating duplicate copies of messages in the folder (which are then handled by the delete-duplicate-messages hook and getting flagged for deletion immediately).

- The fixed version of the code is not doing the right thing either if we insist on the length of the cache vector being = rather than >=.

This needs a closer investigation.

[Uday Reddy (reddyuday)]

Most of this has been handled in Bug 916695.