I think I noticed this before, but neglected to follow up on it. The mbox-append code has code to unblock signals when closing, but never blocks the signals before acquiring the lock in mbox_open_mailbox_append().
At first, I thought this might have been a goof-up during the mx_ops refactoring. So I took another look at https://dev.mutt.org/hg/mutt/rev/2b9d6165b8b7 https://dev.mutt.org/hg/mutt/rev/d18482f6641e and confirmed this was the behavior pre-refactor too. It looks like other places in the mbox code always block signals before locking and unblock after unlocking, but those are mostly confined to a single function. Does this look like an oversite? Or was blocking signals not deemed necessary across the entire open-append operation? -- Kevin J. McCarthy GPG Fingerprint: 8975 A9B3 3AA3 7910 385C 5308 ADEF 7684 8031 6BDA
signature.asc
Description: PGP signature
