Closed Bug 217049 Opened 21 years ago Closed 21 years ago

{inc}width of menus, tables and forms is not displayed correctly

Categories

(Core :: Layout: Tables, defect)

Product:
Core
Component:
Layout: Tables
Version:
Other Branch
Platform:
x86
Windows XP
Type:
defect
Priority:
Not set
Severity:
minor

Tracking

()

Status:
RESOLVED DUPLICATE of bug 215857

People

(Reporter: ht990332, Unassigned)

References

(
URL
)

Details

(Keywords: regression) 

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5b) Gecko/20030822 Mozilla Firebird/0.6.1+
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5b) Gecko/20030822 Mozilla Firebird/0.6.1+

in some webpages such as http://www.flexbeta.net the width of the left forms is
more than it should be, while the width of the right ones is less than it should
be. This error happens around 75% of the time on www.flexbeta.net 

Reproducible: Sometimes

Steps to Reproduce:
1.open http://www.flexbeta.net
2.reload the page, sometimes it is displayed correctly and sometimes when you
refresh it is not.

Actual Results:  
75 % of the time, the width of the forms is not displayed correctly

Expected Results:  
the width of the formashould be displayed correctly
Tthe "main menu" section stretches to the right, pushing everything else off the
screen, it may be easier for the devs to understand.
Summary: width of menus, tables and forms is net displayed correctly → width of menus, tables and forms is not displayed correctly
the left menus are wider than intended so the right ones will be pushed off screen


I can reproduce with Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.5b)
Gecko/20030817 Mozilla Firebird/0.6.1+

I was able to do so by following a link to an article page on the same site.

Also, someone please change the URL field; it has a comma in it.

David
The reason the problem is not 100% reproducable all the time is that the server
apparantly gives (at least) 2 different versions of the page.  I haven't figured
out why it does this, however.

A quick look through the 'good page' and 'bad page' HTML reveals that they both
have markup problems, but I imagine this particular error has something to do
with a bit of code that looks like this in the 'bad page':
<td width="16%" width="400" >
(though the result is neither 400px nor 16% on my screen)

I haven't been able to repro using a local copy of the 'bad page'. :-(

David
Im getting this bug quite often as well.  2 more sites that produce similar
problems (also confirmed by several other users) are

www.adslguide.org.uk
www.ntfs.org

The adslguide one seems to have the problem slightly differently than ntfs and
flexbeta in that text goes off the right of the screen PAST the tables and there
isnt even a scrollbar there to scroll accross to see the text.

screenshots of all three sites available at

www.darkcity.nildram.co.uk/badrender.jpg
www.darkcity.nildram.co.uk/badrender2.jpg
www.darkcity.nildram.co.uk/badrender3.jpg

Its still doing this exactly the same in todays 3rd sept build.  This bug
appeared at some point between 1.5a and 1.5b but i dont know exactly which date,
at least a month ago though.
This bug started with 1.5b (was not in 1.5a) so I suggest that the 1.5b code is
fixed.
Just discovered that when it renders the page wrongly like this, if you press
the back button and then the forward button it will then render the page
perfectly fine (and instantly because it just grabs it straight out the cache i
assume), even though all it has done is load the page from the cache without
downloading a single byte of new information.  This may help track down what
causes it.
Same on mozilla.

Sending to Browser -> Layout: Tables

It doesn't look like it has to do with width specifiers, it's more like the cell
that contains the menu is rendered as being the full width of the table (as if
it was the only column in the table).
Sending to Browser -> Layout: Tables

and this time I mean it!
Component: General → Layout: Tables
Product: Firebird → Browser
Version: unspecified → Trunk
changing component is a lot more effective when you reassign...

might be a dupe of bug 215857 / bug 217590, although this bug worksforme with
linux trunk 2003091105 and I don't see anything on the page to confirm that.
Assignee: blake → table
QA Contact: asa → madhur
bug not fixed yet. Still occurs in build 20030914.
I you press shift + refresh , the page will appear correctly.
It took me a couple of shift refreshes to get it to render correctly at all. 
This is the second site I've seen something like this occur on where a refresh
fixes the problem.

However, would it have anything to do with these lines in the menu column?:
   <td width="16%" width="400" >
    <table cellspacing="0" cellpadding="3" width="100%" border="0">

Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.5b) Gecko/20030903 Firebird/0.6.1+
Comment #4 - comment #8 seem to suggest this is to do with the "on the fly"
rendering of this page as it is downloaded, so it is going to be hard to get a
testcase for this.

Setting Severity: minor, easy workaround - refresh.
Adding Keyword: regression and setting Version: Other Branch, as per comment #6.

I have just tested the pages again and they all WFM (from the screenshot it
looks like http://www.ntfs.org has been redesigned though).
I'm on winXP Gecko/20030913 Firebird/0.6.1+
I just tested on Mozilla Gecko/2003091604 and it worked there too.
Severity: normal → minor
Keywords: regression
Version: Trunk → Other Branch
> Comment #4 - comment #8 seem to suggest this is to do with the "on the fly"
> rendering of this page as it is downloaded, so it is going to be hard to get a
> testcase for this.

We use the term "incremental", not "on the fly".  Thus "{inc}" in the summary.
Summary: width of menus, tables and forms is not displayed correctly → {inc}width of menus, tables and forms is not displayed correctly
bug is still in 1.6a.

I wouldnt say this was minor as its going to stop me using mozilla or fb
completely if its not fixed soon.  Its still there in 23rd sept builds (and ntfs
hasnt had a revamp, by default it uses a different theme than the one in my
screen, which also has the same problem, the theme used doesnt affect it).

I have just gone through the builds and i can confirm that whatever is causing
this was introduced between the nightlies of 29th july and 30th july.  The last
working build that doesnt have this problem was 29th july, the first build that
had this problem (and every build since) was 30th july.  These dates are based
on aebrahims builds (and his 29th build was a late one compiled at 0230AM on
30th, his 30th build was compiled 2000 on 30th according to the timestamps on
the files, so whatever is causing it was introduced at some point in the 17.5
hours in between).

Now that should make this relatively easy to track down and fix, I've provided
proof of the bug, given 3 websites it happens on AND given you the EXACT date
the problem was introduced.....I can do no more.
Well, that suggests that the cause is one of the checkins within the period:

http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=all&branch=HEAD&branchtype=match&dir=mozilla%2Fcontent+mozilla%2Flayout+mozilla%2Fhtmlparser&file=&filetype=match&who=&whotype=match&sortby=Date&hours=2&date=explicit&mindate=2003-07-29&maxdate=2003-07-31&cvsroot=%2Fcvsroot

The most likely one is the one that rewrote the HR element, which fixed a
significant number of bugs.  We're not planning to back that out to fix a bug
that occurs on one site and probably existed before the checkin for HR elements
(but for blocks rather than HRs).

Does the page in question have HR elements?  Does removing them fix the problem?
 Can you make a simplified testcase that shows the problem?  A simplified
testcase makes problems much easier to understand and fix.
I dont know much html, but i checked ntfs and adslguide and neither have a
single <hr> in them, ntfs doesnt even have a <br> so im guessing it cant be that
checkin that caused it.

well i dunno where you get "one site" from cos theres 3 above that i provided
pics of myself.  Id say it occurs on one hell of a lot of sites for those people
that this effects because it happens on 3 sites that i visit, and another 4 or 5
others that ive seen, and i dont exactly visit many websites outside my
bookmarks at all.  I find by far the easiest way to reproduce it for me is on
ntfs.org by clicking links in the different catagories at the top (which reloads
the page if you click a different catagory from the last one you clicked) on
average it seems to be happening about 1 in every 2 times i visit the page or
click the links at the top.

i cant reproduce it locally.  i tried a 'save page as' on ntfs and loaded the
page from disk a few dozen times and didnt get the problem once.  combine that
with the fact that a refresh which loads the page straight from the cache also
doesnt produce the problem, and i think whoever previously said that it was
something to do with the page being rendered wrongly somehow as its being
downloaded was right.

is it possible for someone to make a few builds, removing one or two different
in checkins each one?  then we can find out for certain exactly which checkin
caused it.

personally im wondering if it might be the "fix border-side parsing" checkin as
this has something to do with css and every one of these sites uses a css. 
almost all the other checkins were seemingly to do with view source which i
assume cant possibly have caused this.
Did you test that all the sites mentioned regressed on the same day?  The
problems could be different bugs.
If the pages don't have HR's, then the most likely cause is probably the content
sink changes.

FWIW, I don't see any of the problems mentioned on any of the sites, so it's
hard for me to test.  If there's something I need to do in addition to just
trying to load the URLs, please say so.
i tested all the sites and they all work fine in the same build and are all
broken again in the next.

yes all im doing is loading the pages nothing else and i have no plugins.
> Does the page in question have HR elements?

the flexbeta page is full of them, but they're invisible.  see line 131
(probably reponsible for behavior described in comment 1):

 <table cellspacing="0" cellpadding="3" width="100%" border="0">
  <tr>
   <td bgcolor="#F9FCFF" style="border: #555555 1px solid">
    <img ... /> <b>Main Menu</b>
    <hr align="left" style="width:100%;color:#555555;border:none;height:1px" />
                                  ^^^^

this is a %width specification on an HR in a table column, which is bug 215857.
 the "regression" was from bug 38370, but only because that bug made HR behave
like blocks (blocks had the bug before 38370).

the same html pattern exists on ntfs.org.  I don't see any HR at
www.adslguide.org.uk, but might easily be a different bug.

anyway, I don't actually see the bug on the flexbeta site, but (based on the
description and html) it's a dupe.

*** This bug has been marked as a duplicate of 215857 ***
Status: UNCONFIRMED → RESOLVED
Closed: 21 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.