205147 - textbox doesn't respond when onkeypress="event.stopPropagation"

Status: RESOLVED FIXED

(Core :: DOM: Events, defect)

Description

[Matthew Buckett]

User-Agent:       Mozilla/5.0 (X11; U; Linux ppc; en-US; rv:1.4a) Gecko/20030403
Build Identifier: Mozilla/5.0 (X11; U; Linux ppc; en-US; rv:1.4a) Gecko/20030403

When keypress events are stopped (prevented from bubbling) in a textbox the only
keypresses that seem to have an affect are Enter and Delete. Normal keypresses
are ignored by the textbox. 

This problem only seems to be present in mozilla 1.4, mozilla 1.3 and earlier
seemed to work correctly.

Reproducible: Always

Steps to Reproduce:
1.Try to type in the "stopped" textbox in the example page 


Actual Results:  
No text is displayed in the textbox.

Expected Results:  
Typed text should have been displayed

Comment 1

[Boris Zbarsky [:bzbarsky]]

Sounds like we're treating stopPropagation as being equivalent to
preventDefault; it should not be....

Comment 2

[Andrew Schultz]

this regressed between linux trunk 2003032305 and 2003032405, implicating bug 54035

Comment 3

[saari (gone)]

->bryner for investigation, regression

Comment 4

[Samir Gehani]

adt: nsbeta1+/adt2

Comment 5

[Jesse Ruderman]

The same thing happens with oncontextmenu.  Calling e.stopPropogation() prevents
the context menu from appearing.  I've removed oncontextmenu from the "zap
events" bookmarklet for now so that clicking "zap events" doesn't kill context
menus, but that means "zap events" can't restore context menus when a web site
blocks them.

Comment 6

[jsp]

Attachment: alternate test case

Mozilla apparently also supports IE's cancelBubble property, which is also
affected by this bug.  Attaching alternate test case that demonstrates both.

Is there any hope of this being fixed before 1.4 final now that RC1 is out
there? I understand that the next version of Netscape will be based on 1.4; I'd
hate to have to tell people NOT to use the latest versions of Mozilla,
Netscape, etc, with our site.

Comment 7

[jsp]

Changing OS to All, since it shows up on Linux and Windows.

Requesting 1.4 blocker status, because:
1) this is a regression,
2) it makes some forms unusable,
3) it's difficult to work around, and
4) I'm not sure it's on anyone's radar.

If bug 54035 is truly the cause of this regression, shouldn't it be backed out
until a patch that doesn't cause regressions is available?

Comment 8

[Asa Dotzler [:asa]]

mozilla1.4 shipped. unsetting blocking1.4 request.

Comment 9

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Now, our editor handles key events in system group. So, this bug can be reproduced anymore.