Closed Bug 36080 Opened 26 years ago Closed 26 years ago

[regression] doctype causes wierd behaviour

Categories

(Core :: Layout: Tables, defect, P3)

x86
Windows 98
defect

Tracking

()

VERIFIED WONTFIX

People

(Reporter: doronr, Assigned: rickg)

References

()

Details

Attachments

(3 files)

If you have a doctype defined and a image inside a table, it's height is messed (too big). taking the doctype away makes it work. this can be seen on mozillazine.org attaching testcases. seen on m16 2000041708
Attached file testcase with bug —
Attached file testcase without bug —
hmm... there is another funny thing with the second testcase, first it looks correct and after some seconds, it has the same size(or problem) as the first testcase !?? another thing is does mozilla use "Standard Mode" for HTML 4.0 Transitional or not ? Testet with Build ID 2000041717 on WinNT (this does not occure one day before)
horwath is right - i did not notice that after a few seconds, the bug is back. very peculiar.
This is a feature. If the DOCTYPE contains a SystemID, transitional will be used. In other words, what should happen (although I'm not sure that's exactly what is happening) is that: <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40-971218/loose.dtd"> should trigger strict mode, but <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"> should trigger quirks mode. (If that's not what's happening, it's a bug.) This is being done because: * There aren't that many pages that use a systemID, and generally they're written by hand (and therefore shouldn't need quirks...) * It's what MacIE5 does * It allows authors whose pages only validate to the transitional DTD to still get strict mode.
I think this is not realted to doctype, as the 2nd testcase gets the bug after a few seconds without the doctype element. waiting for the netscape people on this.
*snip * There aren't that many pages that use a systemID, and generally they're written by hand (and therefore shouldn't need quirks...) snip* hmm.... the main reason i´d never use strict mode is the table behavior. then i thought oh perfect then i used this one (this was about 6 month ago, because some of my pages broke in a ugly way) : <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40-971218/loose.dtd"> or this one : <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"> i like the feature switching between strict and quirk mode, but it would be nice if there is a documention or a link in the docu how mozilla reacts. i found it some time ago in a bugreport from dbaron@fas.harvard.edu, but for any other people it is extremly difficult to find. by the way : <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"> forces strict mode in the latest build -> e.g. http://www.ringana.com And this is what happening, and it's a bug...
If you want an image to take up a table cell without space under it, have a rulesuch that any image occupying a table cell gets 'display: block'. But, since this is a bug in the quirks-mode determining code, assigning to rickg. (I'd like to look at this on the weekend, but I can't promise anything...)
Assignee: karnaze → rickg
The parser is acting perfectly well,here. In the first case, it's seeing the "transitional" and HTML 4 specifiers and choose to set the mode to no_quirks. In the second (without the doctype) then we have to choose quirks mode. This appears to me to be behaving according to our spec.
Status: NEW → RESOLVED
Closed: 26 years ago
Resolution: --- → WONTFIX
rickg - thing is, the html woul dmea that the height should be 2 (as the image is defined so). However, it is much bigger than 2 pixels in the first, and becomes bigger in the 2nd after a few secs. this was not happening with m16's from last thursday.
I think the reason the second testcase is changing after a few seconds is that the image is being replaced by alt text when it doesn't load (it's hard to tell what, because it's black-on-black).
Attached file new test case —
dbaron is right = the first 2 do not have the url to the image, so it loads the alt. however, the new testcase has a working image, and with m16, the table is too tall. In the m15 release, this bug does not happen. It shws a image wiuth 2 pixels height. Seems as if a < br > is added to the table. I think this is a valid bug. or am i mistaken? thanks
The code that sets quirks mode is in a little bit of a state of flux right now. There is a good chance that this will change again...
ok, with m16 2000041908, the bug has gone. ;)
Status: RESOLVED → VERIFIED
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: