Closed Bug 71501 Opened 24 years ago Closed 24 years ago

Displaying buglist locks up mac for 5 seconds, then 8 seconds

Categories

(Core :: Networking: HTTP, defect)

PowerPC
Mac System 9.x
defect
Not set
critical

Tracking

()

VERIFIED INVALID
mozilla0.9

People

(Reporter: mikepinkerton, Assigned: gagan)

Details

(Keywords: regression)

- load my buglist mac (G4 450) will lock up for 5 seconds, then display the page, then lock up for 8 more seconds. this is bad.
this happens to a lesser extent displaying today's blocker list (only 3 bugs). there is still is a noticeable lockup.
my build from 3/6 also does this, though the delays seem a little shorter (4 seconds, then 4 seconds).
it's probably the multi-part/replace stuff. I saved my buglist to a file, stripped off the "please wait..." stuff and it loads just fine (no delays) off my local disk. --> gagan.
Component: Layout → Networking: HTTP
really moving to gagan
Assignee: karnaze → gagan
jud anything changed recently in the multipart mixed area?
On my g4, the wait is considerable but only slightly more than the wait in 4.7. Do you really think that this is a tree blocker, I don't.
We don't crash and we end up seeing the bug list OK so I'm downgrading to critical.
Severity: blocker → critical
I am removing smoketest per above comments.
Keywords: smoketest
great, one more reason to use IE for daily work.
Keywords: nsdogfood
MultiMixed stream conversion changed: * date: 2001/02/27 12:33:12; valeski - sr=rpotts. 39987. handling broken servers which don't provide a boundary token in the first block. also protecting against stray parsing which can result in crashes. * date: 2001/03/03 00:59:42; valeski - sr=rpotts. 61160. allowing buffering if we haven't started a sub response yet. Neither of these changes would have the impact pink is describing though. 39987 - doesn't apply to the bugzilla server response type. 61160 - doesn't apply to the bugzilla server response type. I cannot reproduce any "hanging" behavior on my G4 w/ the 3/9 nightly build, or my debug build from a pull at noon today. I submit the query, the busy cursor comes up, the "please wait" page comes up, then the buglist comes in after the query is done (few seconds, but that's expected, all platforms obviously have to wait for the long buglist response to come back). At each point in the process I can mouse, click other things on the desktop, etc. Note: the page doesn't "finish loading" but I think all platforms have that bug. Pink, by "hang" do you mean your machine becomes unresponsive to all events? No mouse cursor movement, no click response? Any chance you have event handling diffs in your tree, or socket transport diffs?
pink: so far no one has been able to duplicate this problem. Any chance you have some changes in your tree as jud suggests? QA: need help reproducing this...
Keywords: qawanted
Target Milestone: --- → mozilla0.9
by 'hang' i mean the menu clock stops ticking (i have it displaying seconds so i can see things like this) and i get the busy watch cursor. the entire machine is trapped, though i can move the mouse. 5 seconds later, things go back to normal (clock ticks, cursor goes back to arrow, etc).
this is an opt install build, so i can't have any changes.
Pink, can you reproduce this in a debug build? If so, could you try getting a nsSocketTransport nspr log?
i'll try an opt and debug build. brade just tried a debug and said she didn't see it. why is my mac the crack-baby one this week? cc'ing more mac folk.
ok, I do see a problem with displaying a bugzilla bug list but I don't see the clock stop like pink does. When I load a bugzilla bug list, the page doesn't seem to ever finish loading. The list itself appears fine but the animation keeps going like it's waiting to get more data. If I view the HTML Source, I don't see the closing tags for body and html.
well guess what. today it doesn't do it. you know, it's probably because bugzilla changed out from under us and is now doing totally different things. i'll close this out as invalid since we can no longer duplicate the same scenario. Crap. I hate it when this happens. There really was a bug there and now it's been masked, ready to nip us right before we ship :(
Status: NEW → RESOLVED
Closed: 24 years ago
Resolution: --- → INVALID
QA - To verify this invalid, lets find some multipart mixed sites that behave like buzilla and hammer on them.
brade: FWIW the bugzilla page not finishing is a known bug -- bug 53956
Marking verified per last comments.
Status: RESOLVED → VERIFIED
Keywords: qawanted
You need to log in before you can comment on or make changes to this bug.