Re: [Mutt] #3000: Deleting >50k messages from IMAP boxes fails with
#3000: Deleting >50k messages from IMAP boxes fails with Fatal error:
---------------------+------------------------------------------------------
Reporter: myon | Owner: brendan
Type: defect | Status: reopened
Priority: minor | Milestone: 1.6
Component: IMAP | Version: 1.5.17
Resolution: | Keywords: patch
---------------------+------------------------------------------------------
Comment(by cameron):
"Second, when the buffer is exceeded, the sender should simply block, not
return a short writes."
If the buffer is nearly full, the write may put in what it can. UNIX
write(2) calls have always been permitted to return short writes. _Is_
there anything watching for the return traffic?
You shouldn't need to "start fetching replies so the server can send
replies and drains your write buffer at the same time" (TCP will happily
send ACKs without outbound data), but if you don't collect the incoming
replies it will only buffer so much inbound data and will block the sender
eventually. If the sender's blocked trying to send to you and not reading
your own sends, you too will block in the same way.
I feel like I describing stuff you already know though...
--
Ticket URL: <http://dev.mutt.org/trac/ticket/3000#comment:30>
Mutt <http://www.mutt.org/>
The Mutt mail user agent