It would be great if support for these proprietary features could be removed from Gecko. Opera encountered one problem so far with not supporting document.width and since Internet Explorer does not have them either it may not be too late to not forever have to support this.
Created attachment 464390 [details] [diff] [review] Patch v1
Did you check that this passes tryserver?
I don't have try access. Tests pass locally, though. Does this need sr?
Comment on attachment 464390 [details] [diff] [review] Patch v1 Wouldn't hurt
Thanks, Dão! Updated https://developer.mozilla.org/en/DOM/document#section_1 https://developer.mozilla.org/en/DOM/document.width https://developer.mozilla.org/en/DOM/document.height https://developer.mozilla.org/en/Firefox_4_for_developers#DOM
This change broke graphs.mozilla.org. A cursory google code search shows that document.height is far from unused, though mostly on aging code paths. I don't see a good reason to take this for Firefox 4. I'm going to back this out.
"Opera had no problems" isn't sufficient research on which basis to remove things, especially without doing a basic search for use. Please do back it out, Sayre, and let's be a little more careful in the future?
Note that "IE has never supported them" was the stronger reason that we felt ok with this. I'm fine with not taking this for firefox 4, however I don't think "graphserver uses it" is a very good reason for keeping a feature in the web platform. Will look more at google code results though.
Ms2ger: Would you mind writing up a patch that adds warnings to the console whenever these properties are used? Probably want to add a flag so that we only warn once per document or some such.
This is not worth focusing on for Firefox 4, imho.
Please back out of mozilla-central default and GECKO20b4_20100817_RELBRANCH
Backed out on default and the b4 relbranch: https://hg.mozilla.org/mozilla-central/rev/9aa39b619a19 https://hg.mozilla.org/mozilla-central/rev/9f434423bdf9
(removing blocking:beta4+ since the issue's no longer in that beta)
The MDC changes also need to be reverted.
Anne's wrong-footing Sicking! What next? Some Googler bamboozling a JS purist into removing __proto__? :-/ /be
The documentation edits need to backed out as well.
Bredan, geez. Jonas expressed interest in removing legacy stuff on the public-html list so I found some and reported a bug with the information I had on Opera and IE. Did I know your internal systems rely on it...
Anne, I kid (mostly :-/). But it is not just our internal systems that use this old stuff. /be
Brendan: It almost is only us using it. Google code search turned up a lot of things that didn't use document_height or other similarly named things, a lot of stuff commented out, and a lot of stuff that falls back to using other properties, or that also depends on layers and thus is already broken. Here are a few things that might be breaking: http://nk-gesture.googlecode.com/svn (nkGestures.js) http://accessext.googlecode.com/svn (treebox_class.js) Some gecko specific thing, also uses XUL git://github.com/cauld/sideline.git (sideline.js) http://konverta4.googlecode.com/svn (functions.js) http://hyperic-hq.svn.sourceforge.net/svnroot/hyperic-hq (footer.js) git://github.com/cauld/sideline.git (ti-sideline-synch-hack.js)
(In reply to comment #21) > > Here are a few things that might be breaking: Removing these is not a terrible idea, but it will take time to research. The fact that one of our sites broke immediately is not a dealbreaker, but it is a bad sign. Thinking more, I can't see how this change improves the Web very much if there's no risk in removing these properties. I'm pretty sure there are much more important things to focus on.
Before removing properties like document.height/.width we could warn in the error console about use of deprecated feature. That approach has worked reasonable well when removing/no-op'ing some Netscapeisms from the event handling code.
Created attachment 467844 [details] [diff] [review] Warning patch v1 (This doesn't apply to trunk yet, due to Mounir's recent changes.)
> Created attachment 467844 [details] [diff] [review] > Warning patch v1 Wouldn't it be better to recommend document.body.clientHeight/clientWidth ? It has better backwards and browser compatibility than document.body.getBoundingClientRect()
Comment on attachment 467844 [details] [diff] [review] Warning patch v1 Though I would also recommend clientWidth/clientHeight.
Comment on attachment 467844 [details] [diff] [review] Warning patch v1 Okay, will fix.
Comment on attachment 467844 [details] [diff] [review] Warning patch v1 Please re-request approval once a patch that's ready to land is available.
Created attachment 468978 [details] [diff] [review] Warning patch v2
Created attachment 469385 [details] [diff] [review] Warning patch (checked in) Thanks.
Comment on attachment 469385 [details] [diff] [review] Warning patch (checked in) http://hg.mozilla.org/mozilla-central/rev/dc76a10a7bdf
Does this really need landing? Looks like it landed. Should it be resolved FIXED?
Yes, the main patch here still needs to land, after we branch. It was backed out: http://hg.mozilla.org/mozilla-central/rev/9aa39b619a19
I was going to land this, but I wasn't sure whether the warning patch should be backed out or not (I think it should).
(In reply to comment #25) > Wouldn't it be better to recommend > document.body.clientHeight/clientWidth ? > > It has better backwards and browser compatibility than > document.body.getBoundingClientRect() In the (admittedly unusual) case where the body element has a border, then clientHeight/clientWidth deliberately excludes that. (Padding is not excluded.) On the other hand, getBoundingClientRect.height/width returns a float...
Updated documents mentioned in Comment 32 Added note to https://developer.mozilla.org/en/Firefox_6_for_developers
Why was this functionality removed? This feature is used in Sun / Oracle ILOM web management interfaces and removal has broken the ability to manage systems using Firefox 6 web browsers and above. Users are then forced to use other browsers. Although Firefox 6 and above support has been fixed in newer ILOM firmware versions by Oracle on some systems, it still means that older historical systems will not work with Firefox 6+. Surely it would have been better to have an about:config setting to re-enable this functionality if needed?