Re: windows IME (bug 3447)

From: Daniel Glassey (danglassey@ntlworld.com)
Date: Thu May 22 2003 - 09:17:14 EDT

  • Next message: Andrew Dunbar: "Re: windows IME (bug 3447)"

    Andrew Dunbar wrote:
    > --- Daniel Glassey <danglassey@ntlworld.com> wrote: >
    > Hiya,
    >
    >>I have a probable fix for this as well (i.e. I now
    >>can type some stuff
    >>in but am not certain if it is working as it is
    >>supposed to since I
    >>don't know cjk). Also, I don't know the current
    >>state of IME's on latin
    >>Win95, or any cjk Win9x (though I do know it is
    >>broken on Win2k+). It
    >>adds another lib dependency but afaik this time it
    >>is a standard lib,
    >>but I still want to check before committing ;). It's
    >>imm32.lib (libimm32.a).
    >
    >
    > Daniel, slightly different code paths are required
    > for IME on ANSI systems and IME on UNICODE systems.
    > The ANSI code will work on UNICODE systems but doesn't
    > allow as much coverage.
    > The old code was known to work with both though
    > finding a CJK Win95/98/ME to test with is the hard
    > part!

    well, the point of this bug is that IME on the pre-unicowspatch build
    _doesn't_ work on Unicode systems.

    Also, I don't know what it was like before but a current (with unicows)
    build doesn't work with the MS Global IME on Win95. Is it supposed to,
    or is it only a CJK Win9x that it is supposed to work with?

    What I do in this fix is catch the WM_IME_COMPOSITION message and handle
    it. I've attached the patch to the bug so you can see what I've done.
    Afaict this is the MS suggested way to do it
    http://msdn.microsoft.com/library/default.asp?url=/library/en-us/intl/ime_5s1e.asp

    > Please see my other mail regarding this library
    > dependency.

    replied to that in the bug.

    Regards,
    Daniel



    This archive was generated by hypermail 2.1.4 : Thu May 22 2003 - 09:33:03 EDT