217886 - frame structure gets lost in builds since 20030826

Status: RESOLVED FIXED

(Core :: Layout: Images, Video, and HTML Frames, defect)

Description

[Peter Möller]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5b) Gecko/20030829
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5b) Gecko/20030829

The hp http://www.mrt-gutesundneues.de/ is designed with 2 frames _menu and
_main. If I click a link in the menu the corresponding html page will not be
shown in the expected frame _main anymore but acting like _top. This means there
are no frames anymore but this page (no _menu frame). Comparing with IE V5.5
will show the expected behaviour. 

Hints: 
- see it since 20030826
- Firebird V0.6.1 works correct

I call it a major bug, because it changes the designed layout of web sites and
should be fixed in V1.5

Reproducible: Always

Steps to Reproduce:
1.
2.
3.

Comment 1

[Shark Daddy (Jonathan L.)]

Indeed. Win2k/20030830.

Comment 2

[Oliver Klee]

Adding regression kw and reassigning to Frames owner & qa.

Comment 3

[Boris Zbarsky [:bzbarsky]]

What's the last nightly build in which this worked correctly?  What's the first
one in which it failed?  There were no checkins on the 25th that should have
affected this in any way.  If people need builds, I temporarily have some Linux
ones going back to Aug 13, 2003 at http://web.mit.edu/bzbarsky/mozbuilds/

Does this work if the frame names do not start with '_' (frame names starting
with '_' are reserved in HTML, and the behavior of using them is undefined)?

Comment 4

[Peter Möller]

Hi Boris,

You got some good kind of instinct ;-) :
Using the frame names without '_' solves the problem.

Still I'm thinking about, if it is a good approach to change a behaviour, that
worked from version V0.9 until the beginning of august 2003. 

Your question concerning the version is not that easy for me. Let's say it like
this:
build 20030802 definitely worked, I 'believe' that 20030822 also did the job.
But maybe analysis is more easy anyway, since we know about '_' now ;-) 

Comment 5

[Mats Palmgren (inactive)]

Bug also occurs in 2003-08-08-22 trunk Linux.

Maybe a regression from bug 215041? (checked in 2003-08-04)

Comment 6

[Andrew Schultz]

regressed between linux trunk 2003080305 and 2003080505

Comment 7

[Mike Connor [:mconnor]]

the site in question uses _Main as the frame name, which works under IE, however
_main behaves in the same way as content.  Because the patch for bug 215041 uses
name.EqualsIgnoreCase("_main") as the check, _Main gets handled the way _main
does, which is not matching IE's behaviour.

hyatt: I'm quite sure that having _main as a case sensitive check will achieve
what we want here without breaking sites using _Main (whether that's correct
behaviour or not, if we're just trying to support IE's behaviour we only need
"_main"

Comment 8

[Boris Zbarsky [:bzbarsky]]

hyatt doesn't read bugmail...

r+sr=me on the change to a case-sensitive check if a comment is added at the
same time explaining why it's done (the attributes involved are all
case-insensitive, so what IE does is a gross HTML spec violation).

Someone will need to actually check the patch in, though -- I have no CVS access
here.

Comment 9

[Christian :Biesinger (don't email me, ping me on IRC)]

if someone attaches a patch here, I can check it in (when it gets approval or
when the tree opens for 1.6a)

Comment 10

[Mike Connor [:mconnor]]

Attachment: patch to make _main a case-sensitive target

This patch makes it a case-sensitive check, thanks to biesi on IRC for the help

Comment 11

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 130852 [details] [diff] [review]
patch to make _main a case-sensitive target

r+sr=bzbarsky if you take out the stray space after "Equals(".

Comment 12

[Mike Connor [:mconnor]]

Attachment: fixing whitespace per bz

I'll get this whitespace thing down sooner or later, thanks bz

Comment 13

[Mike Connor [:mconnor]]

Comment on attachment 130857 [details] [diff] [review]
fixing whitespace per bz

carrying over r+sr,
requesting approval for checkin to 1.5.

Comment 14

[Brendan Eich [:brendan]]

Comment on attachment 130857 [details] [diff] [review]
fixing whitespace per bz

Please get this in for 1.5.

Where is the followup bug for fixing the too-many places that need to know
about these magic names?

/be

Comment 15

[timeless]

checked in

Comment 16

[Mike Connor [:mconnor]]

Fixed on trunk.

Brendan, I filed bug 218254 as a followup.

Comment 17

[Brendan Eich [:brendan]]

mpconnor: thanks.

/be