As noted in bug 82720 ... ------- Additional Comments From sol@netscape.com 2001-07-16 13:49 ------- I'm still seeing this on the branch (commercial) using 2001-07-15-08 on Windows98. Type "quote ibm" in the URL bar, and the resulting stock quote page won't finish loading. The status bar reads: Read C:\PROGRAM FILES\NETSCAPE\NETSCAPE 6\res\arrow.gif ------- Additional Comments From chris hofmann 2001-07-16 14:36 ------- using sol's test case that tries to load http://personalfinance.netscape.com/finance/quotes/quotes.tmpl?symbol=ibm nearly all the content on the page loads but I see the page never finish to load and trobber conintues to throb.. sometimes I see it make it no further than "...conecting to personalfinanace.netscape.com" sometimes "...transfering data from ads.web.aol.com" sometimes "resolving host search.netscape.com" sometimes "Read: C:\..\Netscape 6\res\arrow.gif" branch build from friday Mozilla/5.0 (Windows; U; Win98; en-US; rv:0.9.2) Gecko/20010713 Netscape6/6.1 ----------------------------------------------------------------------------- So, just loading that image directly results in this "infinite throbber". Removing that image from a local copy of that page cures the problem. bbaetz, can you have a quick look at the packets for the HTTP info. Is there anything you can see that's not right?

Okay, so the response headers are (from a packet sniff of the browser loading this image). HTTP/1.0 200 OK Content-Length: 4746 Content-Type: image/gif Expires: Tue, 17 Jul 2001 00:12:29 GMT Also, this 'throbbing forever' does not occur if I uncheck "Enable Keep Alive" in prefs 'Debug->Networking' (with a current trunk build). So, who's doing something wrong: the server? or mozilla?

-> evangalism Whats happening is that the server is responding as an HTTP/1.0 server, and sending a content-length header, but no keep-alive headers. Our behaviour in this case is to ignore the content-length (because some 1.0 servers send incorrect data), and wait for the connection to close. See http://bugzilla.mozilla.org/show_bug.cgi?id=83960 for why we do this (and the rfc justification) However, charts.netscape.com keeps the connection open as if we were doing keep-alives, so we never finish. Keeping the connection open by default would be valid if charts.netscape.com was an http/1.1 server, because then we would expect this to happen. If I change our http version to 1.0 (which doesn't do keep-alives by default), then the server doesn't do keep-alives, and closes the connection, and we work correctly. solution: charts.netscape.com needs to either: - not do keepalives - identify as a 1.1 server, or - parse/understand/respond to the netscape http/1.0 keep-alive extentions which we (and IE) use. It appears to be parsing and understanding the keep-alive stuff correctly, just not responding with the correct headers. Reassigning to default evangalism owner. sol - who am I meant to assign netscape.com evangalism issues to?

Looks like this bug has ended up in the right hands - Bob Clary (who may re-assign to another evangelist if necessary). Bob - this is pretty high priority, since My Sidebar will direct a lot of traffic to the quotes page, and - with the infinite throbber - the experience feels broken.

ok. will contact someone asap.

I have contacted someone at Netscape.com familiar with the personal finance channel and asked for help in contacting the appropriate people about charts.netscape.com. I will update as more information becomes available.

Are we sure this is an evangelism issue. I have seen the throbber continue to throb on many pages, and I do not recall this behavior occuring one month ago. For example - I get the infinite throbber on Microsoft support pages. http://support.microsoft.com/directory/

Well, I have gone through several contacts already. The latest news is that the server is on the east coast and people are trying to locate an administrator. My last contact on this issue was 7/19, so I am giving them more time to respond before trying some of the identified possibilities to contact.

sol: I don't see an infinite throbber on that page. You could be seeing bug 77072, or bug 82720, perhaps?

Bradley: I'm hoping it's bug 82720, but a comment in that bug from John Morrison to me in that bug suggests that the infinite throbber problem is a different problem. FWIW - I'm running 2001-07-20 commercial branch on Win98. Not sure if that explains why we are seeing different behaviors.

sol - your bug isn't 82720, because I reproduced it on a build with the patch for that applied. I'll look into it more once my debug build finishes.

sol - I can't reproduce that bug with debugging turned on, which makes me suspect that its bug 77072, and we're failing on loading a 1x1 gif, or something.

Bradley: You may be right. I found the following in the "charts.netscape.com" page: <img src="/images/pixel.gif" ALT="" WIDTH="1" HEIGHT="1" BORDER="0"> Does this indicate the 1x1 pixel gif problem you were referring to that is the subject of bug 77072?

sol - I don't know, and I won't unless I can reproduce this with debugging on :(

I don't think this is it. I can load the image: http://pfquotes.netscape.com/finance/images/pixel.gif with no problem (i.e., no infinite throbbing).

I ran into this again today in a different context (the page was doing a inline JS image load of one of these images, which meant that onload never fired, and didn't trigger the display of the chart. Subtle variation; same root problem). At any rate, cc: darin just FYI And ping bclary. Any answer from server ops. This would seem to be something that needs to be solved soon, one way or the other.

I spoke to darin about this, and we did some tests. We think we know what IE does, now. It basically appears to read (from the network socket) up to the next boundary of its buffer, and if it could read beyond the size given in the content-length header, it will display that buffer as well. It does this even for 1.1 servers, when not using keep-alives. In other words, for content-length: 1, with a 10K file, it reads 1K (including headers), and displays that. That doesn't change the fact that the server is incorrect, though, but it does explain why IE doesn't ignore the content-length, and can still cope with bug 83960, as well as this test case. jrgm - were you running into this exact image? or a different one?

I was running into the same server/image, except that from the surrounding content on the page, and the apparent symptoms, at first I didn't recognize that. After stripping down the page, I realized that this "new" bug was just this bug. [But it reminded me to ping this bug, since I don't want to have 'our' browser not be able to talk to 'our' stock quotes page when 'we' ship].

i really want to land the fix bbaetz describes... but, i think i'll open a separate bug to cover that fix.

All Evangelism Bugs are now in the Product Tech Evangelism. See bug 86997 for details.

follow up with these sites by 0.9.5

Darin, did you get around to file the new bug? Which one is it? I don't think that this is right in Evangelism. Recently (last months), I saw the infinite throbbing much more often than before. It might be related to this bug. I.e. there might be more servers with this problem.

the "new bug" was bug 89113 and it is now fixed.

-> 0.9.6, need to follow up on these asap

wfm moz trunk 2001 11 14 03/win2k

very nice, fast and throbber stops verify