Closed
Bug 88701
Opened 23 years ago
Closed 22 years ago
Some form pages are loaded twice
Categories
(Core :: Internationalization, defect)
Tracking
()
mozilla1.2alpha
People
(Reporter: beanladen, Assigned: smontagu)
Details
Attachments
(1 file)
(deleted),
image/png
|
Details |
Linux 2.4.5/gcc-2.95/XFree86-4.0.3/0.9.2+
(don't know where this exactly belongs to)
Since 0.9.2, I see form pages loading twice sometimes.
They begin loading, then appear shortly and immediately load again.
(can be seen most easily when clicking through bugzilla bug lists).
This reminds me of the very annoying and time-consuming behaviour
of the netscape 4.7x series, which exhibit this quirk always.
Hope that's not a permanent regression.
Comment 1•23 years ago
|
||
grabow@darkstar.inka.de, we're going to need more information to be able to do
anything about your issue. Can you please take a quick look at
http://www.mozilla.org/quality/bug-writing-guidelines.html for the kinds of
information we need in a bug report. Most useful would be URLs for the sites
where you're seeing this unexpected behavior and descriptions of where in the
layout of the page you see the reflow happen. Thanks for your help in testing
Mozilla and reporting bugs.
Assignee: asa → karnaze
Component: Browser-General → Layout
QA Contact: doronr → petersen
I don't know what else you need here (I know the
requirements, I've read the bugzilla advices several
times...). If you read my report, any necessary information
is given (except the build ID, which bugzilla did
not set this time as usually: 2001063009) ==>
>> (can be seen most easily when clicking through bugzilla bug lists)
I think you know where to find your bugzilla bug lists....
By the way, I saw this problem also when writing >>THIS<<
page (or to be more formal:
http://bugzilla.mozilla.org/show_bug.cgi?id=88701).
>> (don't know where this exactly belongs to)
I'm REALLY not an expert in Mozilla internals, sorry...
But my 20 years of programming experience tell me that
this bug may not be located where it is expected to be.
(I once thought that we germs (unfortunately I'm one
of those) are the most annoying and formal guys, but
you americans are quickly catching up....)
Nevertheless, it's excellent work that you are doing
and I will continue help as best as I can to catch
those remaining quirks in this wonderful product !
grabow: one thing we need to reproduce this bug would be an URL or a testcase,
which show this problem. This would make it easier for us to say, if it is a new
bug and where it belongs to. I sadly can not reproduce it on the bugzilla page
you mentioned (linux 2001070121).
This bug is now in layout, why do you think it does not belong here?
thanks ahead!
Comment 4•23 years ago
|
||
From the description provided, I can't reproduce this issue under Linux Redhat
7.1 with the July 2nd build. Reporter, please check in the latest Linux build.
Checked the last gcc2.95 build today 2001070309.
No change in behaviour. Still THIS page here exibits
the problem (among a couple of other bug reports),
but it's not exactly URL bound.
To be specific, I see the page building up to about the
header of 'Additional Comments', then mysteriously
the Mozilla dragon head appears twice nearly inside
the 'Add CC:' and the 'Component/Status/Resolution' fields,
and shortly after that the page clears and begins to build
up again (seems there's some internal rearrangement of
components going on which is visible to the outside).
Other pages just load and show only a short halt instead of
clearing and building up again, which is really time
consuming for larger pages than this one.
There's NO network activity (reload) during the
second buildup, so the Layout component maybe the
right addressee (I suspected also a caching problem
or something specific to form handling, as this only
applies to form pages). As said, the same effect can
be seen on old Netscape 4.7x series on Solaris 7 for a
wider range of page types.
Please note that I use the gcc-2.95 build WITHOUT Redhat
compatibility libraries, since I have a completely
gcc-2.95 glib221 built 2.4.x linux system (no specific distribution,
but closely resembling a Suse 7.2). The mainstream branch linux builds
crash very quickly and violently on this system, whereas
I've seen no such problems with the gcc-2.95 build.
i remember that I see this behaviour at home with a slow connection (56k).
Perhaps Hyatt's 1.2 seconds painting suppression (bug 77002) makes reproducing
with faster connections so hard, since the page is finnished before the 1.2
seconds are over?
Status: UNCONFIRMED → NEW
Ever confirmed: true
Ah ! Good hint ! I downloaded this page and viewed it locally
==> no problems at all !
Seems an ISDN connection is just too slow.
I think its now on the layout freaks, to decide, if this is really a bug or a
feature. Mozilla is very agressive and tries to render a page, even if it is not
completly available. So the behaviour you described might be wanted in most
cases, even if it looks strange in some situations.
In that case, this bug would be invalid :-/
if someone from the layout department looks at it:
go to a large bug in Bugzilla using a slow connections (56K or ISDN) and look at
the place where all the pulldown menus and input forms comes in later...
Partly agree. I think this should be handed over to the
performance group, maybe this gives a hint how unnessary
page clearing and redrawing can be circumvented.
Comment 10•23 years ago
|
||
I have W2K, Mozilla 0.9.2, no proxy.
I'm not sure if it's just an issue with forms, I'm seeing this behavior on _a
lot_ of pages, e.g. http://www.rpgdot.com or the mozilla bug pages, more often
the larger a page seems to be. And yes, it is only apparent with slow
connections (I have ISDN)...
Now, if Mozilla would merely redraw a page from scratch, making use of the
cache, that would not be a problem, but it [i]reloads[/i] the entire site from
the server. Sometimes downloading a page is almost finished, and suddenly it
would start over. Very annoying.
Arhu
Comment 11•23 years ago
|
||
FWIW, I see something like this with several bugzilla bug pages on Win98 using
build 2001082703.
When the page starts to load and do layout, the form controls appear where they
are supposed to. Then something (reflow?) happens which causes many controls to
disappear (but not the labels). The form controls and labels then appear below
the original form. When I click on a remaining form control (input) in the top
form, the page jumps to and puts focus in the corresponding control in the
bottom form. The page source (using view source) looks to be OK.
I'll attach a screenshot.
Note that some page content is duplicated in the CC (textarea?) box also...don't
know if it's related to this bug or not.
The bad content in the textarea appears after scrolling the page with the scrollbar.
Maybe bug 24397 is related?
Comment 12•23 years ago
|
||
Comment 13•23 years ago
|
||
Linux build 2001082708. I was seeing this with bugzilla bug reports, but with
this version it appears to be fixed.
Comment 14•23 years ago
|
||
*** Bug 99981 has been marked as a duplicate of this bug. ***
Comment 15•23 years ago
|
||
Comment 16•23 years ago
|
||
reassigning to core owner...oh boy this is a doozy. I cant reproduce it at all,
but apparently its low tide or something and i dont have a chicken to sacrifice.
Assignee: karnaze → attinasi
Comment 17•23 years ago
|
||
see also bug 117347 (maybe dup) where the double load has side effect in reflow
and when saving page locally
Reporter | ||
Comment 18•23 years ago
|
||
Maybe a good page to see this effect:
http://aktuelles.t-online.de/nach/pano/ungl/arti/CP/ar-lima.html
(this will probably not live long)
The page reloads very late after it has completed status.
This happens even if you click the reload button after.
Updated•23 years ago
|
Target Milestone: --- → mozilla1.2
Comment 19•22 years ago
|
||
Hi,
I post here to confirm the bug is in Mozilla 1.2a
Also, there seem to be many dups of this bug. Please look at:
Bug #99981 and Bug #77702
How to reproduce:
1. use a tranparent proxy (Squid)
2. run "tail -f access.log | grep html" on your Squid proxy
3. restart Mozilla
4. Go to: www.gamespot.com
5. click on the top of the screen in the "News" link
6. in the top of the page, you find this line:
"Read News: ALL | CONSOLE | PC | XBOX | PS2 | GC | GB "
Click on the "PC" Link
7. the page load twice
8. in your Squid proxy log, you see this:
1032486987.858 3849 192.168.1.1 TCP_MISS/200 46764 GET
http://gamespot.com/gamespot/filters/news/0,10849,6013054,00.html -
DIRECT/206.16.7.15 text/html
1032486996.796 8849 192.168.1.1 TCP_MISS/200 46764 GET
http://gamespot.com/gamespot/filters/news/0,10849,6013054,00.html -
DIRECT/206.16.7.15 text/html
Always reproducible: Yes
How to reproduce again: Follow the above steps. Remember to restart Mozilla.
Notes:
This page has:
- meta:
<meta http-equiv="expires" content="Tue, 20 Aug 1996 01:00:00 GMT">
<meta http-equiv="Pragma" content="no-cache">
<meta NAME="ROBOTS" CONTENT="NOARCHIVE">
- page ends with this line:
<SCRIPT LANGUAGE="JAVASCRIPT"
SRC="http://www.zdnet.com/include/ads/campaigns/pages/spnlinks/includes/gamespot_perm.js">
</SCRIPT>
</body>
</html>
- Only happends in Mozilla. In IE6.0 the page loads only once as reported by
Squid proxy log.
- Tested in Mozilla 1.1Final and 1.2a on WinXP
Comment 20•22 years ago
|
||
This has something to do with my profile.
I created a new profile and can't reproduce the bug anymore.
Then, I restart Mozilla with my old profile and the bug is still
there. So, this has something to do with a corrupted profile.
I have been using this profile since, mmm, don't remember. :-)
It has to have like 1,5 years old and I tryied all the Mozilla
official releases(betas or stable)
Comment 21•22 years ago
|
||
Solved!
I was debugging my pref.js file, deleting lines and trying the test case from
comment #19 and when I deleted this line:
user_pref("intl.charset.detector", "zh_parallel_state_machine");
the bug was solved.
I don't know in which build I got that line in my prefs.js file,
but when I create a new profile in 1.2a, that line is not added in the new
prefs.js file for the new profile.
Can someone check this to see if it works for someone else too?
Comment 22•22 years ago
|
||
another to solve this bug is by selecting the menu:
View->Character Coding->Auto-Detect->(Off)
Comment 23•22 years ago
|
||
Is this bug still reproducible after following sugestion in Comment #22 ?
Auto-Detect->(Off)
Comment 24•22 years ago
|
||
Two things :
- If the targeted page contains a good encoding line (ie <meta
http-equiv="Content-Type" content="text/html; charset=iso-8859-1"> for instance)
there's no problem.
- This behavior exist on post forms too, so the data are posted twice ...
Comment 25•22 years ago
|
||
->Internationalization
Assignee: attinasi → smontagu
Component: Layout → Internationalization
QA Contact: petersen → ylong
Assignee | ||
Comment 26•22 years ago
|
||
*** This bug has been marked as a duplicate of 61363 ***
Status: NEW → RESOLVED
Closed: 22 years ago
Resolution: --- → DUPLICATE
You need to log in
before you can comment on or make changes to this bug.
Description
•