Attachment virtual buffer in mshtml not created if not focussed .doc added by briang1 on 2010-05-15 22:51
Description:
more details and logs

Comment 2 by jteh on 2010-05-19 05:07
In future, it'd be great if you could include details about the issue in the ticket description or comments, rather than attaching it as a Word document. Logs should be attached as text files. Thanks.

Unfortunately, I can't reproduce the issue (despite trying many times) using IE7 and Windows XP, even using the same web page activated via a link on the desktop.

Note that NVDA never forces (or used to force) the focus anywhere in IE. We just respond to the focus events we receive. Even in the log exhibiting the problem, focus is definitely inside the document container, but for some reason, NVDA never sees the focus event for the document itself.

Comment 3 by jteh on 2010-06-17 00:52
Is this improved at all by 5cbec63?

Comment 4 by jteh on 2010-08-19 11:58
Can't reproduce. Until we can, there's nothing we can do to fix this. While annoying, it's possible to work around by moving the focus, so it won't block the release.
Changes:
Milestone changed from 2010.2 to None

Comment 5 by briang1 on 2010-09-20 08:54
The rather strange fix for this has been found in Windows. If you turn on delete temporary internet files when you close IE7, the next time you use IE it has no problems with triggering the reading of the page, etc.

I can only think its either something to do with the delay caused by the save action in a folder with a lot of files in it, or some other not understood reason. If its a drive speed issue then it would be very hard to define it exactly, hence the inability to duplicate.

Quite what it was that changed in between nvda versions to make this important I will leave to you for an idle moment

Comment 6 by briang1 on 2011-06-03 12:34
Just recently, and its hard to say quite which snap started it, this problem has resurfaced even with the temp store turned off and in IE8.
To me, this could be different, or the same, and it looks like as you suspected before, that IE has made its focus somewhere silly, thus the buffers do not get created till you move the focus back to the main page part of the screen. I have a short log fragment below showing this.

INFO - nvda (12:56:17):
Starting NVDA
INFO - core.main (12:56:17):
Config dir: C:\Documents and Settings\brian\Application Data\nvda
DEBUG - core.main (12:56:17):
setting language to Windows
INFO - core.main (12:56:17):
NVDA version main-4386
INFO - core.main (12:56:17):
Using Windows version sys.getwindowsversion(major=5, minor=1, build=2600, platform=2, service_pack='Service Pack 3')
INFO - core.main (12:56:17):
Using Python version 2.7.1 (r271:86832, Nov 27 2010, 18:30:46) v.1500 32 bit (Intel)
INFO - core.main (12:56:17):
Using comtypes version 0.6.2
DEBUG - core.main (12:56:17):
Creating wx application instance
DEBUG - core.main (12:56:21):
Speech Dictionary processing
DEBUG - speechDictHandler.SpeechDict.load (12:56:21):
Loading speech dictionary 'C:\Documents and Settings\brian\Application Data\nvda\speechDicts\default.dic'...
DEBUG - speechDictHandler.SpeechDict.load (12:56:21):
14 loaded records.
DEBUG - speechDictHandler.SpeechDict.load (12:56:21):
Loading speech dictionary 'builtin.dic'...
DEBUG - speechDictHandler.SpeechDict.load (12:56:21):
3 loaded records.
DEBUG - core.main (12:56:21):
Initializing speech
INFO - synthDrivers.espeak.SynthDriver.init (12:56:22):
Using eSpeak version 1.45.03 22.Mar.11
DEBUG - speechDictHandler.SpeechDict.load (12:56:22):
Loading speech dictionary 'C:\Documents and Settings\brian\Application Data\nvda\speechDicts\espeak-english.dic'...
DEBUG - speechDictHandler.SpeechDict.load (12:56:22):
file 'C:\Documents and Settings\brian\Application Data\nvda\speechDicts\espeak-english.dic' not found.
INFO - synthDriverHandler.setSynth (12:56:22):
Loaded synthDriver espeak
INFO - core.main (12:56:22):
Using wx version 2.8.11.0 (msw-unicode)
DEBUG - core.main (12:56:22):
Initializing braille
INFO - braille.initialize (12:56:22):
Using liblouis version 2.3.0
INFO - braille.BrailleHandler.setDisplayByName (12:56:22):
Loaded braille display driver noBraille
DEBUG - core.main (12:56:22):
Initializing NVDAHelper
DEBUG - core.main (12:56:22):
Initializing displayModel
DEBUG - core.main (12:56:22):
Initializing GUI
DEBUG - core.main (12:56:22):
Initializing appModule Handler
DEBUG - core.main (12:56:22):
initializing Java Access Bridge support
DEBUG - core.main (12:56:22):
Initializing winConsole support
DEBUG - core.main (12:56:22):
Initializing UIA support
DEBUG - core.main (12:56:22):
Initializing IAccessible support
DEBUG - core.main (12:56:22):
Initializing input core
DEBUGWARNING - inputCore.InputManager.loadLocaleGestureMap (12:56:22):
No locale gesture map for language en
DEBUGWARNING - inputCore.InputManager.loadUserGestureMap (12:56:22):
No user gesture map
DEBUG - core.main (12:56:22):
Initializing keyboard handler
DEBUG - core.main (12:56:22):
initializing mouse handler
DEBUGWARNING - RPC process 3328 (nvdasd.exe) (12:56:22):
Thread 2144, nvdaHelper\build\x86\remote\ia2Support.cpp, installIA2Support, 81:
Error registering class object, code -2147221008

DEBUGWARNING - RPC process 3328 (nvdasd.exe) (12:56:22):
Thread 2144, nvdaHelper\build\x86\remote\ia2Support.cpp, installIA2Support, 81:
Error registering class object, code -2147221008

DEBUG - core.main (12:56:22):
Initializing global plugin handler
DEBUG - core.main (12:56:22):
starting core pump
DEBUG - core.CorePump.init (12:56:22):
Core pump starting
DEBUG - core.main (12:56:22):
Initializing watchdog
INFO - core.main (12:56:22):
NVDA initialized
DEBUG - core.main (12:56:22):
entering wx application main loop
IO - speech.speak (12:56:24):
Speaking Snapshot Downloader dialog Newest revision: 4386 Installed revision: 4376'
IO - speech.speak (12:56:24):
Speaking revision: edit read only Alt+e'
IO - speech.speak (12:56:25):
Speaking 4376'
IO - inputCore.InputManager.executeGesture (12:56:31):
Input: kb(desktop):shift+tab
DEBUGWARNING - RPC process 3328 (nvdasd.exe) (12:56:31):
Thread 2144, nvdaHelper\build\x86\remote\ia2Support.cpp, installIA2Support, 81:
Error registering class object, code -2147221008

IO - speech.speak (12:56:32):
Speaking revision: edit read only Alt+r'
IO - speech.speak (12:56:32):
Speaking 4386'
IO - inputCore.InputManager.executeGesture (12:56:33):
Input: kb(desktop):shift+tab
DEBUGWARNING - RPC process 3328 (nvdasd.exe) (12:56:33):
Thread 2144, nvdaHelper\build\x86\remote\ia2Support.cpp, installIA2Support, 81:
Error registering class object, code -2147221008

IO - speech.speak (12:56:33):
Speaking button Alt+c'
IO - inputCore.InputManager.executeGesture (12:56:34):
Input: kb(desktop):shift+tab
DEBUGWARNING - RPC process 3328 (nvdasd.exe) (12:56:34):
Thread 2144, nvdaHelper\build\x86\remote\ia2Support.cpp, installIA2Support, 81:
Error registering class object, code -2147221008

IO - speech.speak (12:56:34):
Speaking change log button Alt+v'
IO - inputCore.InputManager.executeGesture (12:56:36):
Input: kb(desktop):space
IO - speech.speak (12:56:36):
Speaking [- speech._speakSpellingGen (12:56:36):
Speaking character u'space'
IO - speech.speak (12:56:38):
Speaking [u'Windows Internet Explorer'](u'pressed']
IO)
IO - speech.speak (12:56:40):
Speaking dialog'
IO - speech.speak (12:56:40):
Speaking Internet Explorer'
IO - speech.speak (12:56:40):
Speaking [- tones.beep (12:56:41):
Beep at pitch 116.271984462, for 40 ms, left volume 50, right volume 50
IO - speech.speak (12:56:41):
Speaking [u'2 percent'](u'pane']
IO)
IO - tones.beep (12:56:42):
Beep at pitch 119.540834878, for 40 ms, left volume 50, right volume 50
IO - speech.speak (12:56:43):
Speaking [- tones.beep (12:56:43):
Beep at pitch 122.901585188, for 40 ms, left volume 50, right volume 50
IO - inputCore.InputManager.executeGesture (12:57:16):
Input: kb(desktop):tab
IO - speech.speak (12:57:16):
Speaking [u'Address edit Alt+D'](u'pane']
IO)
IO - speech.speak (12:57:16):
Speaking http://www.nvda-project.org/log/main?verbose=on&rev=main,4386'
IO - inputCore.InputManager.executeGesture (12:57:26):
Input: kb(desktop):tab
IO - speech.speak (12:57:26):
Speaking Control tool bar'
IO - speech.speak (12:57:26):
Speaking View: websites designed for older browsers will often look better,\r\nand problems such as out-of-place menus, imag button 4 of 7'
IO - speech.speak (12:57:27):
Speaking View: websites designed for older browsers will often look better,\r\nand problems such as out-of-place menus, images, or text will be corrected. tool tip'
IO - inputCore.InputManager.executeGesture (12:57:28):
Input: kb(desktop):tab
IO - speech.speak (12:57:28):
Speaking Command Bar tool bar'
IO - speech.speak (12:57:28):
Speaking button 1 of 1'
IO - inputCore.InputManager.executeGesture (12:57:30):
Input: kb(desktop):tab
IO - speech.speak (12:57:30):
Speaking [- speech.speak (12:57:30):
Speaking [u'Tab Row tab control'](u'pane']
IO)
IO - speech.speak (12:57:30):
Speaking - main (log) tab selected NVDA - main (log)\r\nhttp://www.nvda-project.org/log/main?verbose=on&rev=main,4386'
IO - inputCore.InputManager.executeGesture (12:57:38):
Input: kb(desktop):downArrow
IO - inputCore.InputManager.executeGesture (12:57:38):
Input: kb(desktop):tab
IO - speech.speak (12:57:38):
Speaking Bar tool bar'
IO - speech.speak (12:57:38):
Speaking split button Alt+m 1 of 8'
IO - inputCore.InputManager.executeGesture (12:57:41):
Input: kb(desktop):tab
DEBUG - treeInterceptorHandler.update (12:57:41):
Adding new treeInterceptor to runningTable: <virtualBuffers.MSHTML.MSHTML object at 0x03B9A130>
IO - speech.speak (12:57:41):
Speaking - main (log) document'
IO - speech.speak (12:57:43):
Speaking \r\nSearch: grouping'
IO - speech.speak (12:57:43):
Speaking [- speech.speak (12:57:43):
Speaking [u'banner landmark align center NVDA Logo graphic link'](u'link']
IO)
IO - speech.speak (12:57:43):
Speaking - main (log)'
DEBUG - queueHandler.registerGeneratorObject (12:57:43):
Adding generator 1
IO - speech.speak (12:57:44):
Speaking object at 0x03A62A10>, u'link graphic NVDA Logo', <speech.IndexCommand object at 0x03B917F0>, u'search landmark edit align left button align center Search', <speech.IndexCommand object at 0x03B91A70>, u'link graphic align left Donate', <speech.IndexCommand object at 0x03B91210>, u'navigation landmark list with 7 items align center \u2022 link Home \u2022 link About \u2022 link Download \u2022 link Documentation \u2022 link Recognition \u2022 link Community \u2022 link Development', <speech.IndexCommand object at 0x03B91BB0>, u'out of list main landmark heading level 1 link align left source: link main @ link main,4386 \n', <speech.IndexCommand object at 0x03B9ABF0>, u'Revision Log Mode:', <speech.IndexCommand object at 0x03B9AF90>, u'radio button checked Stop on copy ', <speech.IndexCommand object at 0x03B9AFD0>, u'radio button not checked Follow copies ', <speech.IndexCommand object at 0x03B9AE90>, u'radio button not checked Show only adds and deletes ', <speech.IndexCommand object at 0x03B9ADF0>, u'View log starting at edit main,4386 and back to edit \n', <speech.IndexCommand object at 0x03BAB4D0>, u'Show at most edit 100 revisions per page. \n'
IO - inputCore.InputManager.executeGesture (12:57:58):
Input: kb(desktop):control+alt+r
INFO - core.main (12:57:58):
Exiting
DEBUG - core.main (12:57:58):
Terminating watchdog
DEBUG - core.main (12:57:58):
Terminating GUI
INFO - config.save (12:57:58):
Configuration saved
DEBUG - core.main (12:57:58):
Cleaning up running treeInterceptors
DEBUG - treeInterceptorHandler.killTreeInterceptor (12:57:58):
Killed treeInterceptor: <virtualBuffers.MSHTML.MSHTML object at 0x03B9A130>
DEBUG - core.main (12:57:58):
Terminating global plugin handler
DEBUG - core.main (12:57:58):
Terminating IAccessible support
DEBUG - core.main (12:57:58):
Terminating UIA support
DEBUG - core.main (12:57:58):
Terminating winConsole support
DEBUG - core.main (12:57:58):
Terminating Java Access Bridge support
DEBUG - core.main (12:57:58):
Terminating app module handler
DEBUG - core.main (12:57:58):
Terminating NVDAHelper
DEBUG - core.main (12:57:59):
Terminating keyboard handler
DEBUG - core.main (12:57:59):
Terminating mouse handler
DEBUG - core.main (12:57:59):
Terminating input core
DEBUG - core.main (12:57:59):
Terminating braille
DEBUG - core.main (12:57:59):
Terminating speech
DEBUG - core.main (12:58:00):
core done
INFO - nvda (12:58:00):
NVDA exit

I doubt if its fixable, but just like to update things.
Changes:
Changed title from "Virtual buffers not being created in IE7 and XP on web pages" to "Virtual buffers not being created in IE7 and IE8 and XP on web pages"

Comment 7 by mdcurran on 2011-06-03 22:41
Is this the bug you believed was brought back by 44487c9 or there abouts?
If so I can provide you with a build with this change taken out for you to test.
Up to when the problem came back, do you know if it still occured if you turned off delete temporary internet files on exit anyway? I.e. did a change in NVDA just make it worse so that your work around no longer worked, or did the change in NVDa actually bring the problem back (meaning that for a time there was no problem anymore at all).
For instance, does NVDA 2011.1.1 have the problem even when that feature is turned off?

Comment 8 by briang1 on 2011-06-04 13:37
Right, first of all, it predominates on two machines. its mainly on pages launched from Outlook express emails and the snap downloader, and they need to be quite long by the looks of it. In answer to the questions i can answer then..

Yes it still does it on the release version.
The 'fix' seems to have been more of a quantitive reduction, than a fix. IE it seems that I suppose the less activity going on in the machine there is, the less it happens. However its not the slowest machines that do it most, and I'm suspecting its more to do with data jamup problems due to, say, use of the fsb by video, sound or hard drive to memory or some such. The net result is, I imagine that something arrives too late or not at all somewhere and the focus gets put somewhere off the page itself. I suppose nvda would not actually know the page was there untill it was actually in the active front process so to speak. After all, other windows update and we don't want to hear that when its not where we are working.

The worst machine is a single core running at 2.66 gig with 1.5 gig of memory, but a rather old nvdia card and the fsb speed is 133mhz which seems slow to me.

The hard drive is also quite old but I'd have thought caching would have papered over any problems there.

Anyway, if I make any discoveries I'll update this. Anyone else having problems?

I don't think the removal of that update will prove much apart from the variability of the problem in a given situation on a given machine.

Comment 9 by briang1 on 2011-06-20 15:34
Well its still the same in the 2011.2 beta1 as one would expect, but playing around a bit I think there are two versions of this. Firstly, there is the one where nothing will bring itup at all, not even triggering a menu etc, and the one where one of the following will work. Tab, from some arbitary focus off main window, and triggering a menu bar option or menu. Then cancelling it. There are some occasions when a refresh and a menu or focus even are needed.
As it seemingly is connected to individual machines, and nobody else seems to have had the problem i guess i will have to live with it.

@Brian1Gaff: Could you elaborate on the status of this?

Even though we haven't yet dropped support for Windows XP, it just isn't feasible for us to put resources into fixing XP specific issues any more. Closing as won't fix.