Version: 3.5.4 (using KDE 3.5.4, compiled sources) Compiler: Target: i586-mandriva-linux-gnu OS: Linux (i686) release 2.6.17.13-mm-laptop-3mdvsmp Go to http://www.riteaid.com/stores/locator/ 1) The page shows several boxes into which you can type text information. However, when you type, the text does not appear in the box. 2) Some of the boxes (like Zip code) are not even displayed. Firefox-1.5.0.9 does not have any of these problems.
I get the same problem on that site, although the zip code box is shown. Downloading the page makes it work, so I guess it's some external js or something?
the zip code box is shown here too, KDE 3.5.6.
Created attachment 19562 [details] testcase alas, references an external script. boy, i'm not diving into that hxclient_v3.js JSF-generated monstrosity! :-o in short, if this bug is ever to be identified, reporter would need to provide a cut down version of that script.
KDE-3.5.6 does show the zip code box but does not let text to be typed. The test case in comment 3 does not prevent text from being typed into the box so its not a test case at all. What is the path of the hxclient_v3.js file?
Yeah, ok, then I forgot to mention... In my testcase, as in riteaid.com, it is possible to enter text _immediately_ after the fields are visible. However, after some seconds (seem to vary between 7 and 20) the page snaps (meaning the tab and statusbar flicker a bit, and sometimes the statusbar changes to "loaded" even though it seems to already have loaded earlier) and it's no longer possible to change the contents of the field - as the JS intercepts and "handles" all keypresses in there. Of course, manipulation by mouse (like pasting) works.
oh, and the path is http://www.riteaid.com/.ibmjsfres/hxclient_v3.js
Same problem in kde-3.5.7.
This fixes the typing: --- ecma/kjs_events.cpp (revision 671589) +++ ecma/kjs_events.cpp (working copy) @@ -286,7 +286,8 @@ case TimeStamp: return Number((long unsigned int)event.timeStamp()); // ### long long ? case ReturnValue: // MSIE extension - return Boolean(event.handle()->defaultPrevented()); + // return false == cancel, so this returns the -opposite- of defaultPrevented + return Boolean(!event.handle()->defaultPrevented()); case CancelBubble: // MSIE extension return Boolean(event.handle()->propagationStopped()); default: (But the form doesn't do anything useful afterwards, not sure why)
Oh, never mind that 2nd part, that's just because I had cookies disabled
SVN commit 681861 by orlovich: Properly emulate getting of returnValue. Fixes typing in input boxes on store locator on riteaid.com CCBUG:141230 M +2 -1 kjs_events.cpp --- branches/KDE/3.5/kdelibs/khtml/ecma/kjs_events.cpp #681860:681861 @@ -286,7 +286,8 @@ case TimeStamp: return Number((long unsigned int)event.timeStamp()); // ### long long ? case ReturnValue: // MSIE extension - return Boolean(event.handle()->defaultPrevented()); + // return false == cancel, so this returns the -opposite- of defaultPrevented + return Boolean(!event.handle()->defaultPrevented()); case CancelBubble: // MSIE extension return Boolean(event.handle()->propagationStopped()); default:
SVN commit 681862 by orlovich: Regression test for #141230, get of .returnValue CCBUG:141230 A baseline/events/141230.html-dom AM baseline/events/141230.html-dump.png A baseline/events/141230.html-render A tests/events/141230.html ** branches/khtml/khtmltests-3.5/regression/baseline/events/141230.html-dump.png #property svn:mime-type + image/png
Funnily enough, trunk now allows typing in my testcase, but still not in the riteaid site... :)
Note that I didn't forward-port this to trunk yet.
SVN commit 711658 by orlovich: Forwardport regression tests for #141230 CCBUG:141230 A baseline/events/141230.html-dom branches/khtml/khtmltests-3.5/regression/baseline/events/141230.html-dom#681862 A baseline/events/141230.html-dump.png branches/khtml/khtmltests-3.5/regression/baseline/events/141230.html-dump.png#681862 A baseline/events/141230.html-render branches/khtml/khtmltests-3.5/regression/baseline/events/141230.html-render#681862 A tests/events/141230.html branches/khtml/khtmltests-3.5/regression/tests/events/141230.html#681862 WebSVN link: http://websvn.kde.org/?view=rev&revision=711658
SVN commit 711671 by orlovich: Forwardport fix for #141230 BUG:141230 M +2 -1 kjs_events.cpp WebSVN link: http://websvn.kde.org/?view=rev&revision=711671
Yup now it's fixed in trunk too.