Closed Bug 55256 Opened 26 years ago Closed 19 years ago

hindi/armenian/georgian/tamle keyboard won't work

Categories

(Core :: DOM: Editor, defect, P3)

x86
Windows 2000
defect

Tracking

()

VERIFIED WORKSFORME

People

(Reporter: ftang, Assigned: m_kato)

Details

(Keywords: access, intl, topembed-, Whiteboard: [t2])

Attachments

(2 files)

on Window 2000. 1. Install Hindi, Armerian, Georgian, or Tamel keyboard by using control panel, Regional Setting 2. open Composer 3. switch your keyboard layout to any of these 4. start typing actual result got question mark for every keystroke expect result got character of those languages. I think the problem is we register ourself as ANSI application instead of unicode application and MS does not define an ACP for those new languages they support in Win2K.
m_kato@ga2.so-net.ne.jp- can you help ?
Status: NEW → ASSIGNED
ftang, that's right. Since Notepad and Wordpad is Unicode application (using EditboxW / Rich20W window class, it is unicode window class), it will not occur... But I think that this fix nedd many codes since Windows 9x cannot use Unicode window class. Roy, please see the following URL. http://msdn.microsoft.com/library/psdk/winbase/ime_371h.htm
Roy, you have no time. please assign to me. I found fix codes...
Attached patch a patch — — Splinter Review
Per Kato-san's request, I am asssigning this bug to him .
Assignee: yokoyama → m_kato
Status: ASSIGNED → NEW
Status: NEW → ASSIGNED
I'll super-review this for you, sorry about the delay. This patch involves quite a bit of code duplication to handle the Unicode/ANSI cases. To avoid most of this duplication, how about creating static methods in nsToolkit for each of the Windows APIs where you need either the wide or the ANSI version, e.g. nsToolkit::PeekMessageN nsToolkit::DispatchMessageN nsToolkit::UnregisterClassN nsToolkit::CreateWindowExN etc, and implement these methods to take the wide parameters and convert them down to ANSI when we're running on Win9x. I think that would reduce the code duplication and simplify the patch a lot. For example, we would only need WindowClassW(). Also, did you really want to check this in? -#if defined(DEBUG_ftang) -//#define KE_DEBUG -//#define DEBUG_IME -//#define DEBUG_IME2 -//#define DEBUG_KBSTATE -#endif +//#if defined(DEBUG_ftang) +#define KE_DEBUG +#define DEBUG_IME +#define DEBUG_IME2 +#define DEBUG_KBSTATE +//#endif and stuff like this? -#define NS_IMM_SETCONVERSIONSTATUS(hIMC, lpfdwConversion, lpfdwSentence) \ +#define NS_IMM_SETCONVERSIONSTATUS(hIMC, lpfdwConversion, lpfdwSentence, bRet) \
A fix code like your comment will occur performance down since PeekMessage and DispatchMesage always call by message loop. So I don't agree agree. And debug code is my mistake:-). Of cource, this codes does not check in... Also, fix of IME macro is since this macro is broken...
There will probably be a very small slowdown, but I think it will be so small you will not be able to detect it. Unless we find a real performance problem, we should use the simpler code.
check in
Resolving this bbug as fixed per m_kato. Changing QA contact to teruko for now.
Status: ASSIGNED → RESOLVED
Closed: 25 years ago
QA Contact: sujay → teruko
Resolution: --- → FIXED
This created a major regression --- bug 63243. I think we need to make sure that plugin windows always use ASCII and not Unicode.
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
Do plugin windows ever have to display non-ASCII strings, e.g. Latin 1 accents, Chinese, Japanese, etc. If so, can that be taken care of with the ASCII assumption?
Makoto Kato, could you back your changes out unitl you find a way to make them work with plugin windows? Thanks!
Raised the severity to "critical". Peter, if this is a blocker, please re-adjust accordingly.
Severity: normal → critical
Plugins are already using ASCII windows. If they carry on using ASCII windows, then nothing is going to get worse. They can certainly display non-ASCII characters even using an ASCII window; Mozilla's been doing that for a long time. Even if we wanted plugins to use Unicode windows, we can't because that breaks them.
there is fix code in bug 63243. mark as FIXED
Status: REOPENED → RESOLVED
Closed: 25 years ago → 25 years ago
Resolution: --- → FIXED
This bug will be FIXED whn an acceptable patch has been checked in that doesn't cause unacceptable regressions. It is not FIXED yet. Here is the reference to your proposed patch on top of your original patch: http://bugzilla.mozilla.org/showattachment.cgi?attach_id=21241 This does not fix the real problem. The real problem is that plugins expect their window to be an ANSI window. This is not a bug, it is part of the implicit contract between Mozilla and the plugins. We can't change that contract. The Get/SetWindowLongA/W problem is just one of the possible things that could break if we violate this contract. Patching in a special work-around for this one case is not a good idea. We need to revise your original patch to make sure that plugin windows are always ANSI. Then we need to test thoroughly with lots of different plugins to make sure we didn't introduce any regressions. And then we can get new reviews and check the whole thing in.
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
Changed QA contact to ylong@netscape.com.
QA Contact: teruko → ylong
Keywords: access, intl
Keywords: nsbeta1
Keywords: topembed
minusing. are we missing something? are these patches still relevant?
Keywords: topembed → topembed-
Whiteboard: [t2]
i18n triage team: nsbeta1- Frank believes this is fixed anyways.
Keywords: nsbeta1 → nsbeta1-
This bug was already fixed at past years since Windows builds supprots Unicode window class.
Status: REOPENED → RESOLVED
Closed: 25 years ago → 19 years ago
Resolution: --- → WORKSFORME
Status: RESOLVED → VERIFIED
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: