When accessing to main gentoo's bugs webpage with Opera 8.01 and trying to put in some search terms or a bug number and clicking "Show" (or pressing Enter key), it does not search. This is, it does not give any output, error, or something. It simply does nothing. Reproducible: Always Steps to Reproduce: Actual Results: It does not search. Expected Results: It should search.
I can't reproduce this. Clicking on show or pressing the Enter key are both working fine here. How often did you try it? Maybe it was just a problem with your internet connection or an overload of the Bugzilla server?
mmmm, weird, today it runs, but I can assure that the past week it didn't ran. There were more people with this people. Perhaps someone can confirm this. Thanks.
It's normal situation with Opera. What can i say, i write "ALL firefox" into the box in Firefox and Opera and in FF i get results, but in Opera no, just loading forever (and javascript is turned on). And i'm not surprised that you can't reproduce it. For some people it works, for some just not. It's just like with bug causing opera to segfault at pages loaded full of stuff, mostly flash banners: it often crashed, i've heard about it from more than 20 people, but no one is able to reproduce it, because when opera is fired up again it works perfect with that page. Piece of closed source junk. I just use firefox instead, and this bug could go to "can't reproduce". Cheers. :] btw, there's some new opera version around, not in Portage yet, worth checking if it gives any results also. :]
(In reply to comment #3) > and this bug could go to "can't reproduce". Cheers. :] I agree.
Happens also with Opera 7; but Opera is innocent; problem didn't exist until recent bugzilla upgrade. Choose "Tools > Quick Preferences > Identify as Opera"; click "Show". Everything works. Choose "Tools > Quick Preferences > Identify as Mozilla 5.0"; click "Show". Wait forever.
Works perfect for me also. Thanks for solution. But still, what's wrong on the bugzilla side?
I don't know, maybe some javascript thing. Bartron, you told me about the browser identification thing; do you have more information?
Well I don't think it's javascript... if you type the query string directly into the address bar, opera "hangs" before it gets a chance to download any external scripts (and the only embedded script is an empty function...). It seems `buglist.cgi' performs differently if the only difference is `User-Agent: '...
Ok, here's the difference... I think the problem is `Content-Type: multipart/x-mixed-replace' if `User-Agent' is Moz5. * User-Agent: Opera/7.54 (X11; Linux i686; U) [en] HTTP/1.1 200 OK Date: Wed, 03 Aug 2005 00:13:39 GMT Server: Apache/2.0.54 (Gentoo/Linux) Content-disposition: inline; filename=bugs-2005-08-03.html Set-Cookie: xxxxx Set-Cookie: xxxxx Vary: Accept-Encoding,User-Agent Content-Encoding: gzip Keep-Alive: timeout=15, max=100 Connection: Keep-Alive Transfer-Encoding: chunked Content-Type: text/html; charset=ISO-8859-1 * User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; X11; Linux i686) Opera 7.54 [en] HTTP/1.1 200 OK Date: Wed, 03 Aug 2005 00:14:04 GMT Server: Apache/2.0.54 (Gentoo/Linux) Content-disposition: inline; filename=bugs-2005-08-03.html Set-Cookie: xxxxx Set-Cookie: xxxxx Vary: Accept-Encoding,User-Agent Content-Encoding: gzip Keep-Alive: timeout=15, max=100 Connection: Keep-Alive Transfer-Encoding: chunked Content-Type: text/html; charset=ISO-8859-1 * User-Agent: Mozilla/5.0 (X11; Linux i686; U) Opera 7.54 [en] HTTP/1.1 200 OK Date: Wed, 03 Aug 2005 00:14:27 GMT Server: Apache/2.0.54 (Gentoo/Linux) Content-disposition: inline; filename="bugs-2005-08-03.html" Vary: Accept-Encoding,User-Agent Content-Encoding: gzip Keep-Alive: timeout=15, max=100 Connection: Keep-Alive Transfer-Encoding: chunked Content-Type: multipart/x-mixed-replace;boundary="------- =_aaaaaaaaaa0"
So the problem is bugzilla? (what does multipart/x-mixed-replace mean?)
No, I think the problem is opera. `multipart/x-mixed-replace' is composed just like any other multipart content... `x-mixed-replace' means all contained blocks are transmitted one after another, and each new content block replaces the previous one once it becomes available (google: SERVER-PUSH). Bugzilla uses this technology to first send a very short page ("Bugzilla is pondering your qery" ... "Please stand by") that is later replaced with the query results. The problem as I see it is that opera seems to have problems if said multipart wrapper is compressed (testcase below)... so a change in bugzilla may have triggered it, but if the Mozilla way of handling it is standards conformant, it's clearly a bug in Opera (more or less, there may be a good reason why bugzilla hands out different content to Opera browsers...) The testcase (see next 3 attachments)... `content.gz' is the raw data sent by bugzilla in response to a query asking for all bugs that changed within the last day (including original gzip compression). `test1.cgi' is a small cgi script that sends `content.gz' uncompressed... this displays correctly in Opera 7.54, Konqeror, Mozilla 4, and Mozilla 5. `test2.cgi' is another small cgi script that sends the same data with `Content-Encoding: gzip', just like bugzilla does... this irritates Opera and Mozilla 4, but works with the other two. If anyone wants to confirm this with Opera 8, just dump all three files into the same `cgi-bin' dir and open `test1.cgi' and `test2.cgi'.
Created attachment 65029 [details] content.gz
Created attachment 65030 [details] test1.cgi
Created attachment 65031 [details] test2.cgi
what about opera-8.5?
still a problem?
I had absolutely no problem in doing my day to day dev-work - in both opera 8.51 and in 9-preview. So for me it's not an issue anymore.
For me, was solved since comment #2, and, of course, with current opera. Feel free to close and solve it. Feel free to close it. Thanks for your work.
nice -> closed
