OSDN Git Service
Mesar Hameed [Thu, 1 May 2014 16:21:07 +0000 (18:21 +0200)]
Merge branch 'l10n'
Mesar Hameed [Thu, 1 May 2014 16:05:16 +0000 (18:05 +0200)]
L10n updates for: zh_CN
From translation svn Revision: 15267
Authors:
vgjh2005@gmail.com
Mesar Hameed [Thu, 1 May 2014 15:08:14 +0000 (17:08 +0200)]
L10n updates for: de
From translation svn Revision: 15263
Authors:
Bernd Dorer <bernd_dorer@yahoo.de>
David Parduhn <xkill85@gmx.net>
Rene Linke <demetry@gmx.de>
Mesar Hameed [Thu, 1 May 2014 13:21:40 +0000 (15:21 +0200)]
L10n updates for: ne
From translation svn Revision: 15260
Authors:
him Prasad Gautam <drishtibachak@gmail.com>
Mesar Hameed [Thu, 1 May 2014 13:21:22 +0000 (15:21 +0200)]
L10n updates for: fa
From translation svn Revision: 15260
Authors:
Ali Aslani <aslani.ali@gmail.com>
Mesar Hameed [Thu, 1 May 2014 13:21:47 +0000 (15:21 +0200)]
L10n updates for: ru
From translation svn Revision: 15260
Authors:
ruslan <ru2020slan@yandex.ru>
Vadim <eye0@rambler.ru>
Mesar Hameed [Thu, 1 May 2014 13:21:44 +0000 (15:21 +0200)]
L10n updates for: pt_BR
From translation svn Revision: 15260
Authors:
Cleverson Casarin Uliana <clcaul@live.com>
Marlin Rodrigues <marlincgrodrigues@yahoo.com.br>
Mesar Hameed [Thu, 1 May 2014 13:21:34 +0000 (15:21 +0200)]
L10n updates for: it
From translation svn Revision: 15260
Authors:
Simone Dal Maso <simone.dalmaso@gmail.com>
Mesar Hameed [Thu, 1 May 2014 13:21:32 +0000 (15:21 +0200)]
L10n updates for: hu
From translation svn Revision: 15260
Authors:
Aron OcsvAri <oaron@nvda.hu>
Mesar Hameed [Thu, 1 May 2014 13:21:37 +0000 (15:21 +0200)]
L10n updates for: ko
From translation svn Revision: 15260
Authors:
Joseph Lee <joseph.lee22590@gmail.com>
Chang-Hoan Jang <462356@gmail.com>
Mesar Hameed [Thu, 1 May 2014 13:21:41 +0000 (15:21 +0200)]
L10n updates for: nl
From translation svn Revision: 15260
Authors:
Bram Duvigneau <bram@bramd.nl>
Bart Simons <bart@bartsimons.be>
A Campen <a.campen@wxs.nl>
Leonard de Ruijter <mail@leonardder.nl>
Mesar Hameed [Thu, 1 May 2014 13:21:52 +0000 (15:21 +0200)]
L10n updates for: ta
From translation svn Revision: 15260
Authors:
Dinakar T.D. <td.dinkar@gmail.com>
Mesar Hameed [Thu, 1 May 2014 13:21:56 +0000 (15:21 +0200)]
L10n updates for: vi
From translation svn Revision: 15260
Authors:
Nguyen Van Dung <dungnv1984@gmail.com>
Mesar Hameed [Thu, 1 May 2014 13:21:24 +0000 (15:21 +0200)]
L10n updates for: fr
From translation svn Revision: 15260
Authors:
Michel such <michel.such@free.fr>
Patrick ZAJDA <patrick@zajda.fr>
Remy Ruiz <remyruiz@gmail.com>
Bachir Benanou <ben_bach@yahoo.fr>
Mesar Hameed [Thu, 1 May 2014 13:21:36 +0000 (15:21 +0200)]
L10n updates for: ja
From translation svn Revision: 15260
Authors:
Takuya Nishimoto <nishimotz@gmail.com>
Mesar Hameed [Thu, 1 May 2014 13:21:29 +0000 (15:21 +0200)]
L10n updates for: gl
From translation svn Revision: 15260
Authors:
Juan C. buno <oprisniki@gmail.com>
Mesar Hameed [Thu, 1 May 2014 13:21:23 +0000 (15:21 +0200)]
L10n updates for: fi
From translation svn Revision: 15260
Authors:
Jani Kinnunen <jani.kinnunen@wippies.fi>
Mesar Hameed [Thu, 1 May 2014 13:21:20 +0000 (15:21 +0200)]
L10n updates for: es
From translation svn Revision: 15260
Authors:
Juan C. buno <oprisniki@gmail.com>
Mesar Hameed [Thu, 1 May 2014 13:21:19 +0000 (15:21 +0200)]
L10n updates for: el
From translation svn Revision: 15260
Authors:
access@e-rhetor.com
Mesar Hameed [Thu, 1 May 2014 13:21:11 +0000 (15:21 +0200)]
L10n updates for: ar
From translation svn Revision: 15260
Authors:
Fatma Mehanna <fatma.mehanna@gmail.com>
Shaimaa Ibrahim <shamse1@gmail.com>
Mesar Hameed [Thu, 1 May 2014 13:21:10 +0000 (15:21 +0200)]
L10n updates for: an
From translation svn Revision: 15260
Authors:
Jorge Perez <jorgtum@gmail.com>
Mesar Hameed [Thu, 1 May 2014 13:21:08 +0000 (15:21 +0200)]
L10n updates for: am
From translation svn Revision: 15260
Authors:
Dr. Tamru E. Belay <g.braille@sympatico.ca>
James Teh [Thu, 1 May 2014 06:42:21 +0000 (16:42 +1000)]
Minor fixes to What's New.
James Teh [Thu, 1 May 2014 06:22:51 +0000 (16:22 +1000)]
NVDA now recovers in more cases when switching away from applications that stop responding.
Fixes #3825.
James Teh [Tue, 11 Feb 2014 06:06:12 +0000 (16:06 +1000)]
cancellableSendMessageTimeout: When sending a message from a thread other than the main one, even though we don't use a background thread, still send in short attempts and check for cancellation.
UI Automation threads send messages to potentially unresponsive windows. This allows us to check for cancellation, though because cancellation is based on the main thread being alive, messages might not be cancelled if the main thread recovers before the thread in question. However, more importantly, sending in short attempts allows us to get around the issue where SMTO_ABORTIFHUNG doesn't abort if the window hangs while the operation is in progress instead of before.
James Teh [Mon, 10 Feb 2014 08:54:30 +0000 (18:54 +1000)]
There's no reason for bgSendMessageData to be dynamically allocated any more, nor is the traget thread id needed. isActive can now be moved inside bgSendMessageData.
James Teh [Mon, 10 Feb 2014 08:20:41 +0000 (18:20 +1000)]
cancellableSendMessageTimeout: Don't ever return early due to cancellation while the background SendMessage is still running, as this can cause crashes in the remote process.
Sometimes, the caller allocates memory in the remote process, sends a message which depends on it and frees it as soon as the SendMessage* call returns. If the background SendMessageTimeout is still running, the message might reach the remote process and it will then use the freed memory, thus causing crashes or heap corruption.
To fix this, the background thread attempts to send the message with 400ms timeouts, checking for cancellation after each attempt, similar to what the main thread did previously. The main thread just waits for the background thread to finish handling the message, regardless of whether it was cancelled. For the most part, this takes us back to what we had before the background thread was implemented, except that we don't get blocked in SendMessageTimeout for longer than we should.
This will hopefully fix crashes in Outlook Express and other applications.
Re #3859.
James Teh [Tue, 4 Feb 2014 11:49:35 +0000 (21:49 +1000)]
Remove a pointless variable.
James Teh [Tue, 4 Feb 2014 11:18:08 +0000 (21:18 +1000)]
Update some copyright headers.
James Teh [Tue, 4 Feb 2014 05:50:42 +0000 (15:50 +1000)]
cancellableExecute: Use cancelCallEvent rather than spinning.
James Teh [Tue, 4 Feb 2014 05:34:45 +0000 (15:34 +1000)]
watchdog: Cache cancelCallEvent so we don't have to keep fetching it.
James Teh [Tue, 4 Feb 2014 02:58:20 +0000 (12:58 +1000)]
cancellableSendMessageTimeout: Rather than continually cancelling individual calls with an auto reset event, just set it once when cancellation starts and reset it when the core is alive.
Previously, it was possible that the first message after the core recovered would be cancelled if there wasn't a message after the last recovery attempt.
Also, this way, we don't bother to send any messages at all once cancellation begins, which is more efficient.
James Teh [Mon, 3 Feb 2014 22:36:23 +0000 (08:36 +1000)]
cancellableSendMessageTimeout: Don't pump if SMTO_BLOCK is specified.
James Teh [Mon, 3 Feb 2014 09:15:15 +0000 (09:15 +0000)]
cancellableSendMessageTimeout: Fix nasty memory corruption if a message is ever sent to HWND NULL, which we sometimes do in our SysListView32 code.
We use HWND NULL to signal to the background thread that it should die, so when a message was sent to HWND NULL, the background thread was dying and deleting its data block, but the main thread was still using it, thus resulting in memory corruption. Furthermore, subsequent messages would simply have been dropped until cancellation. This could be seen regularly when pressing backspace in Windows Explorer in Windows XP, but was probably occurring elsewhere too.
James Teh [Mon, 3 Feb 2014 08:25:07 +0000 (08:25 +0000)]
cancellableSendMessageTimeout: The result variable passed to SendMessageTimeout should be DWORD_PTR, not DWORD.
This should make no difference for NVDA since it is always 32 bit, but best to be correct.
James Teh [Mon, 3 Feb 2014 05:59:12 +0000 (05:59 +0000)]
SendMessage hook: Initialise the result variable to 0 so an uninitialised value isn't returned on error.
James Teh [Fri, 31 Jan 2014 08:01:42 +0000 (08:01 +0000)]
cancellableSendMessageTimeout can't handle reentrancy, so abort early for reentrant calls.
As far as I know, there shouldn't be any reentrant calls to SendMessage* from oleacc or UIAutomationCore in our main thread, but I'm not absolutely certain, so it's best to be safe.
It should be possible to make this code reentrant if we discover this is a problem.
James Teh [Fri, 31 Jan 2014 07:44:24 +0000 (07:44 +0000)]
MSDN says you can't safely read from an STL object while another thread is writing. I doubt this would cause crashes, but acquire the lock just to be safe.
James Teh [Wed, 29 Jan 2014 02:25:09 +0000 (02:25 +0000)]
cancellableSendMessageTimeout: cosmetic: Use a LockableObject instead of directly using a CRITICAL_SECTION. Also, rename/reorganise/comment for readability.
James Teh [Wed, 29 Jan 2014 02:07:44 +0000 (02:07 +0000)]
If message cancellation was already requested by the time cancellableSendMessage starts, just return before doing anything.
James Teh [Wed, 29 Jan 2014 02:01:01 +0000 (02:01 +0000)]
cancellableSendMessage: Pass the data pointer to the thread proc, as the global variable might be null if the message gets cancelled before the thread starts executing, causing nasty crashes.
James Teh [Tue, 28 Jan 2014 10:49:25 +0000 (10:49 +0000)]
cancellableSendMessageTimeout: Place a lock around the unresponsiveThreads set, as writing to STL containers isn't thread safe and was causing crashes.
James Teh [Tue, 28 Jan 2014 09:03:57 +0000 (09:03 +0000)]
cancellableSendMessageTimeout: SendMessageTimeout can block until the window responds in some cases despite the timeout, so use a background thread to send messages.
Michael Curran [Wed, 30 Apr 2014 23:02:55 +0000 (09:02 +1000)]
Merge branch 't1846'. Fixes #1846
Michael Curran [Wed, 30 Apr 2014 22:58:23 +0000 (08:58 +1000)]
Merge branch 't2199'. Fixes #2199
Michael Curran [Wed, 30 Apr 2014 22:55:41 +0000 (08:55 +1000)]
Merge branch 't3481'. Fixes #3481
James Teh [Wed, 30 Apr 2014 01:00:09 +0000 (11:00 +1000)]
Remove extras/NVDA CD, which is extremely outdated and no longer useful.
James Teh [Tue, 29 Apr 2014 07:19:30 +0000 (17:19 +1000)]
In browse mode in Mozilla applications, using quick navigation commands to move to the next or previous button or form field now includes toggle buttons as expected.
Fixes #4098.
James Teh [Mon, 14 Apr 2014 04:29:17 +0000 (14:29 +1000)]
In Microsoft Office 2010 Options dialogs, the labels of combo boxes are reported correctly.
These seem to be exposed correctly by MSAA, but not by UIA. This suggests the UIA implementation is once again broken, so I've disabled it for controls in these dialogs. Hopefully, this won't break something else.
Fixes #4056.
James Teh [Tue, 29 Apr 2014 06:27:58 +0000 (16:27 +1000)]
Bah. Forgot the What's New entry.
Siddharth [Mon, 23 Dec 2013 10:02:13 +0000 (15:32 +0530)]
In Microsoft PowerPoint with speaking of typed words enabled, characters erased with backspace are no longer announced as part of the typed word.
The typedCharacter event isn't fired for backspace, so we have to fake it.
Fixes #3231.
James Teh [Mon, 28 Apr 2014 07:23:29 +0000 (17:23 +1000)]
In browse mode, radio and check menu items are reported as controls instead of just clickable text. Also, NVDA no longer incorrectly switches from focus mode to browse mode when a radio or check menu item is focused.
Re #4092.
Michael Curran [Sun, 27 Apr 2014 23:04:13 +0000 (09:04 +1000)]
Merge branch 't4069'. Fixes #4069
Mesar Hameed [Sat, 26 Apr 2014 15:16:33 +0000 (17:16 +0200)]
en/{userGuide.t2t,changes.t2t}: remove witespace on otherwise empty lines.
Thanks to Patrick Zajda
Michael Curran [Tue, 22 Apr 2014 05:23:30 +0000 (15:23 +1000)]
Merge branch 't4043'. Fixes #4043
Michael Curran [Tue, 22 Apr 2014 05:19:01 +0000 (15:19 +1000)]
Merge branch 't4031'. Fixes #4031
James Teh [Sat, 19 Apr 2014 01:04:16 +0000 (11:04 +1000)]
Reduce spinning in NVDA core and watchdog.
Fixes #3801.
Michael Curran [Wed, 16 Apr 2014 01:00:41 +0000 (11:00 +1000)]
Reverting a change from December 2009 commit
6c53a29d where IAccessible NVDAObject's name property was forced to contain content from IAccessibleText if there was no caret present.
This was supposed to allow focusable labels in Thunderbird to read nicer. However, this caused a long-standing issue where the content of alerts in web browsers would speak twice, and Thunderbird now seems to oddly expose a caret on the labels thus disabling the code change anyway.
Removing this code may possibly cause some other regressions as its very old. However, if regressions are found they should be handled specificly for each case, as clearly this code was too general.
Should be a fix re #3481. May affect re #2442, #2745
Mesar Hameed [Tue, 15 Apr 2014 08:10:28 +0000 (10:10 +0200)]
l10n updates
From translationsvn rev 14937.
Michael Curran [Tue, 15 Apr 2014 05:58:12 +0000 (15:58 +1000)]
speech.speakTextInfo: Stop accidentally reporting controlFields that may slightly change in virtualBuffers (name change, change of child count) by Using a uniqueID value in controlField dictionaries for comparison if it exists rather than compairing the entire dictionary. VirtualBuffers now expose uniqueID generated from the docHandle and ID of the node.
Michael Curran [Tue, 15 Apr 2014 02:16:14 +0000 (12:16 +1000)]
MSHTML VBufBackend: force an invalidated subtree to be re-rendered with no delay if it happens to be inside a live region. to speed up live region reporting even further
Michael Curran [Tue, 15 Apr 2014 02:05:57 +0000 (12:05 +1000)]
MshtmlVBufStorage_controlFieldNode_t: add a reportLiveText method which will speak any given text, as long as it is not entirely whitespace. reportLiveAddition and postProcessLiveRegion both use this method.
Michael Curran [Tue, 15 Apr 2014 01:49:51 +0000 (11:49 +1000)]
Support live regions in MSHTML documents, dialogs and applications by Useing MSHTML virtualBuffers for documents, dialogs and applications, but force focus mode for things that are not documents (e.g. dialogs and applications).
Specifically: move the check out of shouldCreateTreeInterceptor into MSHTML virtualBuffer's constructor and force focus mode if needed.
Michael Curran [Tue, 15 Apr 2014 01:41:55 +0000 (11:41 +1000)]
VBufBackends: decrease the render timeout from 250 ms to 100 ms. Mostly needed to make live region reporting faster.
Michael Curran [Mon, 14 Apr 2014 23:27:59 +0000 (09:27 +1000)]
Support for ARIA live regions in MSHTML, implemented via the MSHTML VBufBackend.
* fillVBuf now takes an oldNode argument, which when called from render, is the original node being re-rendered, but is null for all other recursive calls.
* fillVBuf takes a atomicNodes argument by reference, which is created by render. This set is for temporary tracking of aria-atomic nodes that should be announced when hit or at the end of the render, if a live region update occured within an atomic region.
* MshtmlVBufStorage_controlFieldNode_t: now contains various aria live region properties to track if the node is a live region, what kind of updates are relevant (text, additions), if its busy, and if its part of an atomic region.
* MshtmlVBufStorage_controlFieldNode_t: has a preProcessLiveRegion method, which is called by fillVBuf when adding a node to the buffer. This method updates all of the ARIA live regions properties on the node, either calculating their values from HTML attributes if they exist, or copying them from their parent (in this buffer or the original buffer).
* fillVBuf: takes an inNewSubtree argument, which is false when called from render, but is set to true on recursive calls if the current node was not in the old buffer (i.e. its ID does not point to a node in the old buffer).
* MshtmlVBufStorage_controlFieldNode_t: has a reportLiveAddition method which collects text content and announces it with nvdaController_speakText.
* MshtmlVBufStorage_controlFieldNode_t: has a postProcessLiveRegion method, which is called at the end of fillVBuf if this fillVBuf was called with inNewSubtree set to false (i.e. its not in a new subtree, or its just the top node in a new subtree). This method is responsible for checking if this node is part of a live region and should be reported. It may call reportLiveAddition to report the node if its new, or it may compaire child text with the old node's child text to see if text should be reported and call nvdaController_speakText to report the text. However if the node is part of an atomic region, that atomic region is added to the atomicNodes set for alater reporting.
* FillVBuf: after postProcessLiveRegion is done, atomicNodes is checked to see if the current node is in the set. If so its removed, and has its reportLiveAddition called, thus having the atomic region reported.
* Render: after fillVbuf is called, any remaining nodes in atomicNodes are fetched and reportLiveAddition is called on each, thus reporting any atomic regions in the ancestors of the render, referenced by live region updates within the render.
Michael Curran [Mon, 14 Apr 2014 02:53:01 +0000 (12:53 +1000)]
Merge branch 't4031' into mshtmlLiveRegions_fillVBuf
Michael Curran [Fri, 11 Apr 2014 06:09:26 +0000 (16:09 +1000)]
Don't announce the document itself when entering a browse mode document after the first time, when coming up from a non-rendered descendant. Still announce when coming in from the outside of the subtree.
Michael Curran [Thu, 10 Apr 2014 22:53:10 +0000 (08:53 +1000)]
NVDAHelperRemote: export all nvdaController functions so that they can be easily used by virtualBuffer backends and other in-process dlls related to nvdaHelperRemote.
Michael Curran [Mon, 7 Apr 2014 05:11:02 +0000 (15:11 +1000)]
MSHTML VBufBackend: only skip text nodes if the parent node is hidden due to style, not role="presentation".
James Teh [Wed, 2 Apr 2014 05:08:31 +0000 (15:08 +1000)]
navigatorObject_currentDimensions script: Fix exceptions when errors occur.
When we report an error, let's actually return too. :)
Fixes #4044.
James Teh [Wed, 2 Apr 2014 04:30:52 +0000 (14:30 +1000)]
Merge branch 'master' into t3801
James Teh [Wed, 2 Apr 2014 04:24:10 +0000 (14:24 +1000)]
watchdog: Use core.mainThreadId.
James Teh [Wed, 2 Apr 2014 04:23:42 +0000 (14:23 +1000)]
core.requestPump: optimisation: If called in the main thread, don't bother using CallAfter to start the timer, as this is only needed for other threads.
Mesar Hameed [Tue, 1 Apr 2014 08:03:58 +0000 (10:03 +0200)]
l10n updates
From translation svn rev 14779.
Michael Curran [Tue, 1 Apr 2014 06:14:27 +0000 (16:14 +1000)]
Added an option to keyboard settings dialog which allows to disable handling of key input from other applications. Users typing Vietnamese using Unikey should turn this option off.
James Teh [Tue, 4 Mar 2014 08:27:39 +0000 (18:27 +1000)]
eventHandler: Fix race conditions which could leave NVDA in a state where it constantly reported there were pending events when there weren't, causing weird problems such as menu items not being reported.
Queued events are always executed in the main thread, but they can be queued from any thread. Incrementing/decrementing the pending event counts is not atomic, so place a lock around updates to the count data.
Also, when queuing, update the counts before actually queuing the function. Otherwise, if queuing from a background thread, the queued function might run before the counts were updated.
Fixes #3939.
Mesar Hameed [Sat, 22 Mar 2014 21:47:27 +0000 (22:47 +0100)]
l10n updated german nvda.po to hopefully fix non-utf8 problems.
From translation svn rev 14742.
Mesar Hameed [Wed, 19 Mar 2014 18:18:33 +0000 (19:18 +0100)]
l10n updates:
from translation svn 14717.
James Teh [Tue, 18 Mar 2014 18:38:15 +0000 (11:38 -0700)]
In browse mode, labelled regions are now treated as landmarks.
Fixes #3741.
James Teh [Tue, 18 Mar 2014 17:59:41 +0000 (10:59 -0700)]
In browse mode, labels on landmarks are now reported. They are also included in the Elements List dialog.
Fixes #1195.
James Teh [Tue, 18 Mar 2014 17:42:06 +0000 (10:42 -0700)]
Gecko vbuf backend: Fix rendering of links, buttons, etc. when the label is overridden (e.g. aria-label or title) and the element is invisible but its content is visible.
When an element has a width/height of 0, we still consider it visible if it has children. This should still apply even when we prevent rendering of children because of explicit-name.
Fixes #3991.
James Teh [Mon, 10 Mar 2014 04:54:09 +0000 (14:54 +1000)]
In browse mode, the labels of controls such as links and buttons are now rendered correctly where the label has been overridden by the author for accessibility purposes (specifically, using aria-label or aria-labelledby).
Fixes #1354.
Michael Curran [Mon, 10 Mar 2014 01:19:01 +0000 (11:19 +1000)]
IA2LiveRegions: make sure to free the IAccessible2 once finished with it. Fixes a small memory leak.
Michael Curran [Mon, 10 Mar 2014 01:12:00 +0000 (11:12 +1000)]
NVDA appModule's event_foreground now correctly calls nextHandler. This ensures that speech will be canceled before any NVDA dialog is announced.
Mesar Hameed [Fri, 7 Mar 2014 21:42:58 +0000 (22:42 +0100)]
Merge branch 'rc'
Mesar Hameed [Fri, 7 Mar 2014 21:35:57 +0000 (22:35 +0100)]
l10n:
fr - Fixed an important interface problem, an unmatched { broke the input gesture dialog
Thanks Michel Such.
James Teh [Fri, 7 Mar 2014 06:19:40 +0000 (16:19 +1000)]
Merge branch 'rc'
James Teh [Thu, 13 Feb 2014 17:38:27 +0000 (03:38 +1000)]
Whitespace objects are no longer included in simple review or reported when the focus moves, as they are purely for layout purposes.
Fixes #3839.
James Teh [Fri, 7 Mar 2014 01:44:46 +0000 (11:44 +1000)]
Update eSpeak to 1.48.04 due to an important fix for Dutch.
Michael Curran [Thu, 6 Mar 2014 22:35:56 +0000 (08:35 +1000)]
Merge branch 't3900'. Fixes #3900
Michael Curran [Wed, 5 Mar 2014 22:37:13 +0000 (08:37 +1000)]
Merge branch 't3888'. Fixes #3888
Michael Curran [Wed, 5 Mar 2014 22:34:22 +0000 (08:34 +1000)]
Merge branch 'rc'
Michael Curran [Wed, 5 Mar 2014 06:00:58 +0000 (16:00 +1000)]
Update changes file to note eSpeak 1.48.03
Michael Curran [Wed, 5 Mar 2014 04:32:05 +0000 (14:32 +1000)]
Merge branch 't3860' into rc. Fixes #3860
Michael Curran [Wed, 5 Mar 2014 04:31:49 +0000 (14:31 +1000)]
eSpeak: disable whole-program optimization as this was causing some distortion and worbling on certain Klatt voices when compiled with msvc2012
Michael Curran [Wed, 5 Mar 2014 01:42:17 +0000 (11:42 +1000)]
Upgrade eSpeak to 1.48.03 to fix badly compiled phonemedata Re #3860
James Teh [Mon, 3 Mar 2014 06:48:32 +0000 (16:48 +1000)]
Fix reporting of labels for landmarks in MSHTML documents.
The name attribute is now set if there is a name wherever the name wouldn't usually be rendered as the content. Previously, it was only set in very specific cases, which didn't cover landmarks.
Mesar Hameed [Sun, 2 Mar 2014 13:07:48 +0000 (14:07 +0100)]
l10n:
From translation svn 14518.
Michael Curran [Thu, 27 Feb 2014 04:45:02 +0000 (14:45 +1000)]
Merge branch 'master' into t3888
Michael Curran [Thu, 27 Feb 2014 04:32:21 +0000 (14:32 +1000)]
Merge branch 't3874'. Fixes #3874