GNOME Bugzilla – Bug 414762
[blocked] Up/Down Arrow don't always work in Firefox
Last modified: 2010-04-02 15:02:38 UTC
position: absolute interferes with keyboard navigation within page content. In particular Control Home/End do not always work -- or if they work, they do not always take you to the beginning/end of the page content. Similarly, Up/Down Arrow cannot be relied upon to navigate among an entire page's contents when position: absolute is used. See this Firefox bug: https://bugzilla.mozilla.org/show_bug.cgi?id=372624
Will, do you think we should try to script around this, or block it against the bug I filed against Firefox?
I think we should keep hammering on the Firefox folks to fix their bugs. :-)
Removing target milestone from [blocked] bugs. We have little control over them, so we're better off letting priority and severity be our guide for poking the related components.
Going through my bugs to see where I'm at with things. This one has a target milestone of 2.21.90. The reason we're blocked is busted caret navigation (in particular, problems surrounding floats and position:absolute). What should I be trying to achieve w.r.t. this bug for 2.21.90, clever hacking or gentle poking? :-)
> What should I be trying to achieve w.r.t. this bug for 2.21.90, clever hacking > or gentle poking? :-) I prodded for you. I'd like to see the FF stuff get fixed at the source if possible. If the FF folks cannot fix it, and this particular bug happens very frequently on a large number of pages, and we can work around it, then we should try. But, let's see what they have to say first. Thanks!
Created attachment 102649 [details] [review] proposed patch > I prodded for you. Nobody expects the Spanish Inquisition.... ;-) > I'd like to see the FF stuff get fixed at the source if possible. Agreed. > If the FF folks cannot fix it, I don't believe that for a moment. > and this particular bug happens very frequently on a large number of pages, Between broken caret navigation and CSS positioning preventing it, I'd say it happens on more pages than not, and users on the list have mentioned it a number of times. > and we can work around it, then we should try. Attached. Because we're taking over Control+Home and Control+End, we've got keybindings, and hence new strings. But this is an easy one so I suggest we get it in for 2.21.5. > But, let's see what they have to say first. I've not heard a response other than fixing caret navigation is hard. :-( Mike please test.
I think the patch looks great (thanks!). If it tests well with Mike, I say commit and make sure you send out a string change announcement. Thanks!
Created attachment 102701 [details] [review] handle bogus last objects Mike IMed me to let me know that going to the end wasn't working on overstock.com. I took a look. I hate poorly-coded pages and bogus hierarchies.... <smile> getLastObject() returns the very last object in the document frame. This object should have no children. Therefore, it should have text where we can place our caret. What's different in this patch is that getBottomOfFile() looks for that text and if it doesn't find it, it assumes bogusity and looks to the previous object. Mike please test. Thanks!
After testing the patch to 506360 I'm happy with this patch as well.
Thanks Mike. Patch committed. Can't quite move this to pending I guess since we still have broken caret navigation issues that we're tracking. :-( But I can take the "control home/end" bit off now. ;-) Will: I've will probably have several string announcements to make so I'll do them all at once. But it's noted down. :-)
Changing the target to FUTURE as the Control+Home/End patch was part of 2.21.5 and we're blocked w.r.t. the Up/Down Arrow issues. (This is not the new line nav bug, but a tracking bug related to position:absolute and floats interfering with Mozilla/Gecko caret navigation.)
The purpose of keeping "blocked" bugs open is to track bugs in other modules which we cannot work around within Orca and which we have some expectation of eventually seeing a fix. In this case of this particular bug, we've worked around many of the problems and will do our best to work around others as we discover them. Plus, I see no indication that this issue will ever be resolved on the Gecko side of things. :-( In addition, as specific problems related to line navigation have been discovered, we've opened new bugs in b.g.o to address them. For these reasons, I'm closing this tracking bug out as it is serving no real purpose.