On Tue, 29 Jun 2004, Roland Hill wrote:
>Hi Andreas,
>* Andreas Aardal Hanssen <[EMAIL PROTECTED]> [29-06-04 06:37]:
>> .sent/cur/1084056953.9923_337.rrl03:2,S
>Couldn't find any messages with this ID, however I did find this:
>[EMAIL PROTECTED]:~/Maildir/.sent/cur> ls -al 1088418091.16326_409.rrl03:2,S
>-rw-------    1 roland   users    373968431 2003-10-20 18:29
>1088418091.16326_409.rrl03:2,S
>I have moved this message to another location and fetching the /.sent
>folder now seems to be okay.

Try moving the message in again. If this makes the IMAP server fail again,
then we have a concrete way to find the exact bug. Binc does handle huge
emails quite fine with no extra memory usage. I tested with a 1GB email -
no problem, although it did take some time to parse. If you are able to
reproduce your problem, then I hope we can cooperate off-list to find the
point where Binc caves in.

Andy :-)

PS: Hehe, actually I did a profile of the parser using this 1GB test email
using calltree/valgrind and kcachegrind, and found a way to make the
parser approximately 30% faster ;-). On my machine anyway; and I didn't
account for how much of the time was spend in IO operations so the actual
parser may have gained more than this. I submitted it to 1.3 (the change
is too big for 1.2).

PPS: calltree/valgrind and kcachegrind are _invaluable_ tools for C++
programmers who enjoy optimizing code. Just _look_ at those sexy
screenshots. I warmly recommend these tools to _every single_ C and C++
programmer out there:

http://kcachegrind.sourceforge.net/cgi-bin/show.cgi
http://valgrind.kde.org/

--
Andreas Aardal Hanssen   | http://www.andreas.hanssen.name/gpg
Author of Binc IMAP      |  "It is better not to do something
http://www.bincimap.org/ |        than to do it poorly."

Reply via email to