See problem description in bug 451957 and bug 296453 comment 46. My inbox has 200,000 msgs, most of them spam and marked as Junk. Since I switched to sort by (primary criteria) Junk and (secondard criteria) subject (i.e. click on subject column first, then on Junk column once or twice, so that Junk appears at the end), I observed an excessive RAM usage, and CPU usage when opening the folder. Specifically, startup takes 2 minutes (!) until the first window appears, and uses 700 MB RAM at that point. If I click on a (small) folder and back on inbox, RAM usage grows to 1.3 GB and stays there. Since I switched back to (via the menu) View | Sort By | Order Received and Descending (i.e. newest arrivals at the top), RAM usage dropped significantly again, it's about 300 MB after switching folders back and forth. I have not tried other sort criterias / combinations.
sorting on junk is indeed bad - took ~8 seconds vs 1 second to sort subject during which time the UI is blocked. I'd be surprised if this isn't related to a threading performance bug, like bug 226730 (sorting on read status also seems slow, but not as bad as sorting on junk) ... so backend? In addition to being slow, at least the first two times I sorted again on junk - after being sorted on date or subject - the peak memory jumped: 70 meg on the first sort and 40 meg on the second. the starting point, when sorted on subject was thunderbird using 70 meg. I tested a 16k message local folder using trunk on XP. Folder has no messages marked junk, i.e. a couple dozen messages are non-junk, and the vast majority are undetermined. from Ben's bug 451957, comment 0 is based on : * build 20080721 (~alpha 1) * IMAP on Cyrus server on local private network * 200,000 msgs in Inbox, as said * Almost everything spam (due to bug bug 296453), which means subjects often repeated, and From varying a lot and not in addressbook (if relevant for bug 243631).
Component: Mail Window Front End → Folder and Message Lists
OS: Linux → All
QA Contact: front-end → folders-message-lists
You need to log in before you can comment on or make changes to this bug.