137126 - innerHTML escapes <> everywhere

Status: RESOLVED WORKSFORME

(Core :: DOM: Core & HTML, defect, P5)

Description

[Anders Larsson]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:0.9.9) Gecko/20020311
BuildID:    20020311 and 20020412

very much like entry #60619
The thing is that I've failed utterly to reproduce the problem outside of our
environment.
For me the problem is serious because if i user innerHTML on any element <>
chars get converted to &lt; and &gt;.
To update divs is a pain now.

I get these errors when assigning innerHTML to divs in a frame totally generated
by javascript. I have been unable to mimic this behaviour in any way. It works
in IE and on server generated pages.

Reproducible: Couldn't Reproduce
Steps to Reproduce:
This is the frustrating part.
If I try to lift the code out of our dev environment it works.
What i do is in the description, create string of type "<table> etc..."

Actual Results:  My divs get filled with html-as-text in format "&lt;table&gt;
etc..."

Expected Results:  unconverted html

Comment 1

[Boris Zbarsky [:bzbarsky]]

Without an actual testcase that reproduces this bug, there is not much that can
be done. I  have been unable to reproduce it... please provide a test url or
upload a file using
http://bugzilla.mozilla.org/attachment.cgi?bugid=137126&action=enter that
reproduces the problem.

Comment 2

[Boris Zbarsky [:bzbarsky]]

Anders, any luck with that testcase?  Could you save the entire web page the
problem happens on (using "web page, complete" as the save format) and attach to
this bug?

Comment 3

[Anders Larsson]

Sorry

I could not reproduce this with a local version of my troublesome web page (so
maybe it's not too serious).
I rewrote the script using nodes instead of innerHTML (much better in all
aspects). That was a great learning experience so now I'm happy again.

I was only having problems with this on pages created with
"frame.document.open();frame.document.write(html);frame.document.close()"-style
dhtml and since that is kind of an obsolete way of doing things i don't think
people will hav any problems. 

Only my 0.02 $ but I wouldn't give this to much prioroty.

Comment 4

[Fabian Guisset]

Marking NEW, we will look at it again if/when we have a testcase

Comment 5

[Hixie (not reading bugmail)]

worksforme. we need testcases.