[SCIM] gtk IM & reset ...

[Zhe Su]

Hi,
  So, which input method engine are you using? scim gtk immodule
itself should have no problem. The code is ok.

Regards
James Su


On Mon, 13 Dec 2004 15:54:05 +0000, michael meeks <mmeeks at novell.com> wrote:
> Hi guys,
> 
>         So - I'm having some acutely painful experience with SCIM & it's
> (apparent) problems with it's implementation of gtk_im_context_reset:
> 
> extras/gtk2_immodule/gtkimcontextscim.cxx (gtk_im_context_scim_reset):
> 
> static void
> gtk_im_context_scim_reset (GtkIMContext *context)
> {
>     SCIM_DEBUG_FRONTEND(1) << "gtk_im_context_scim_reset...\n";
> 
>     GtkIMContextSCIM *context_scim = GTK_IM_CONTEXT_SCIM (context);
> 
>     if (context_scim && context_scim->impl && context_scim ==
> _focused_ic) {
>         panel_prepare_transaction (context_scim);
>         _backend->reset (context_scim->impl->id);
>         panel_send_request (context_scim);
>     }
> }
> 
>         Contrast this to:
> 
> gtk+/modules/input/gtkimcontextxim.cxx (gtk_im_context_xim_reset):
> 
>         Which explicitely emits a 'Preedit End' type sequence: ie. a commit
> followed by a 'preedit_changed' to a 0 length string.
> 
>         Worse - than just not emitting the signals to commit / empty the
> pre-edit - it appears, that the _backend->reset doesn't reset the remote
> IM's pre-edit buffer; hence the client/server are in a confused state
> wrt. what is in a pre-edit buffer etc.
> 
>         This causes myriad problems; eg. on 'random mouse click' (cursor
> movement) OO.o will attempt to first reset the IM / commit the current
> pre-edit. While the pre-edit is committed - it appears to stay there -
> and gets extended by subsequent, key-strokes (in a new place); and this
> is just one example of highly sub-optimal behavior ;-)
> 
>         Am I mis-understanding something ? or is this broken ? also - it
> appears the download page for scim is defunct:
> 
>         The links from here
>                  http://www.freedesktop.org/wiki/Software_2fScimDownload
> 
>         ugly <br> tags aside, point to inaccessible URIs - which made reading
> your code somewhat more painful than necessary :-)
> 
>         Thanks,
> 
>                 Michael.
> 
> --
>  michael.meeks at novell.com  <><, Pseudo Engineer, itinerant idiot
> 
> _______________________________________________
> scim mailing list
> scim at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/scim
>