Bug 206261

ok, are we agreed that we all see this...

a) input into kedit works ?
b) input into OOo *and* gedit does not work ?
c) cut and paste from kedit into OOo *and* gedit works ?

Is there an opened bug against some gnome component on this as well, likely a
common input problem with OOo and e.g. gedit.

> a) input into kedit works ?
> b) input into OOo *and* gedit does not work ?
> c) cut and paste from kedit into OOo *and* gedit works ?

Kedit - yes - kdeutils-3.5.4-3.fc6
OOo - new findig - writer crashes on typing "kR" -
openoffice.org-writer-2.0.4-3.1  and libicu-3.6-2
Gedit - does not works - pango-1.14.3-2.fc6 and scim-sinhala-0.2.0-1.fc6
CnP from gedit to kedit works.

> CnP from gedit to kedit works.

CnP from kedit to gedit works.

CnP from kedit to oowriter also works

dudes, I haven't had time to invest in debugging this yet. But it is *so* going
to  be something to do with the generic sinhala input mechanism for gtk apps and
not something specific to OOo.

RE: Comment #5

Tried on gedit, it has the same *wrong* behviour as oowriter. At the moment, it
is leaning towards a pango issue, Bug 204922. Should icu be updated after the
pango is fixed?

I don't know what you're so sure that the problem lies in the pango and icu
renderers when cut and paste from kedit works, and raw code input works :-)

I suspect instead that the IM is somehow not delivering the characters to
gedit/OOo correctly. And I know there was a change to the sinhala input engine
recently so that's an added suspicion

so...

a) Here's a very interesting little fact, it works perfectly fine on x86_64 (!)
for both gedit and OOo (so possibly some timing or event ordering issue?) when I
input "kii"
b) if on i386 I revert scim-sinhala to 0.1.0-3.fc6 and restart scim, then it
works fine on i386 for both gedit and OOo for "kii"

I'll debug this, I suspect something like...

a) k
    emit unicode
b) i
    emit unicode
c) i
    emit backspace
    emit new different unicode

but that backspace gets delivered after the third unicode commit, or something
of that nature

Created attachment 137072 [details]
yeah, example patch

This patch is total crack, and not to be used. It just illustrates the problem,
which is that the backspace for the second unicode char is processed by gtk
after the commit of the third unicode character, so we end up with the 2nd
unicode remaining, not removed and replaced. 

scim-bridge needs to do what the scim gtk component does and try and use
g_signal_emit_by_name for the backspace keypress, e.g. see thread of
http://sourceforge.net/mailarchive/forum.php?forum_id=43684&max_rows=25&style=nested&viewmonth=200504

*** Bug 204922 has been marked as a duplicate of this bug. ***

*** Bug 206108 has been marked as a duplicate of this bug. ***

"ii", "oo", "uu", "dH" are 1st modifier problems while "kRoo" are 2nd modifier
problems. Hence as per Tyronne, this bug is not duplicate of bug 203911

The problem is the difference how forwarding key event is handled between
scim-gtkimm and scim-bridge-gtkimm.
The former handle the event immediately if possible, but the latter doesn't.
Okay, this bug must be fixed in the next release.

Thanks, Dairiki.  I see a change related to this has been committed in cvs.

Dairiki-san, I tested current scim-bridge cvs head with scim-sinhala-0.2.0
(scim-sinhala-trans cvs) and it doesn't seem to be quite right yet:

Now Inputting "oo" gives "ඔ" and then "oඔ" (not "ඕ").
At the same time the following output comes from gedit:

An unknown error occurred at scim_bridge_client_handle_key_event ()
An IOException at filter_key_event ()
An IOException occurred at scim_bridge_client_imcontext_filter_key_event ()
The message is recieved in a wrong context: key_event_handled

*** Bug 207110 has been marked as a duplicate of this bug. ***

*** Bug 206244 has been marked as a duplicate of this bug. ***

I've applied another patch on the cvs latest.
I've confirmed that typing "oo" gives "ඔ" and then "ඕ".
Is it okay?

Wonderful, seems to be fixed now in cvs. :-)
Thank you, Dairiki-san!

I backported the patch in scim-bridge-0.4.5-3.fc6.

This seems to be fixed in  rawhide.

Package version 
scim-bridge-0.4.9-1.fc7
scim-bridge-gtk-0.4.9-1.fc7
fonts-sinhala-0.2.1-1

Steps to reproduce
1. Activate scim-sinhala.
2. in any editor or oowriter.
3. Type the keysequence "oo" 

Result
Sinhala Letter OOyanna OD95, which is the expected result in Comment #1. As the
issue has been fixed, Moving it to Resolved.

(In reply to comment #22)
> This seems to be fixed in  rawhide.
> 
> Package version 
> scim-bridge-0.4.9-1.fc7
> scim-bridge-gtk-0.4.9-1.fc7
> fonts-sinhala-0.2.1-1
> 
> Steps to reproduce
> 1. Active scim-sinhala
> 2. in any editor or oowriter
> 3. Type in the "oo" 
> 
> Result
> Gives me  Sinhala Letter Oyanna OD94
Scratch/Ignore this..happend thr a Mid Air collison ;-)

Created attachment 145415 [details]
Actual result :: Wrong OOyaana combinations

The is was the previous OOyana combination, the Actual error which were
previously occuring

Created attachment 145416 [details]
Expected Result

This is the correct expected result, which is working in the as mentioned in
Comment #24

So all is working correctly in rawhide now yes ? and no further action is being
called for.

Created attachment 145956 [details]
Input of the sinihala comments

o Selecting the scim-sinhala, key sequence
o Input "oo"
o gives you sinhala letter OOyana ඕ(U+0D95)
o Attaching the screenshot of gedit as evidence.

Does it currently work correctly, or does it currently not work correctly ?

If it doesn't work, then reopen and I'll look into fixing it. 
If it does work, then I don't think we need to keep adding comments to this bug :-)

No need to worry, it is all good I think.

In reply to Comment #29

It works :-)