Description of problem: with m17n-db-sinhala and scim-sinhala package, there are 3 layouts >sinhala >wijesekera >samanala with sinhala/wijesekera, there is problem with input some characters, but same characters are work properly with 'samanala'. C 'n' P is also working from gedit Version-Release number of selected component (if applicable): scim-1.4.4-25.fc6 scim-sinhala-0.1.0-3.fc6 m17n-db-sinhala-1.3.3-14.fc6 How reproducible: Steps to Reproduce: 1. open kedit 2. select scim-sinhala 3. type ko 4. with wijesekera 5. type fdl 6. repeat step with gedit to see correct output Actual results: wrong rendering in kedit/Openoffice Expected results: Result should be like gedit Additional info:
Aman, Can you cut and paste the characters and see if OOo or kedit renders correctly? If not then it is not about input method but it is a rendering problem.
Yes, copy 'n' paste is also working from gedit
My question was: Does cut and paste works with OOo and kedit?
correct in kedit, but not correct in Openoffice (copy Paste from gedit)
Hopefully this will be fixed when we upgrade to scim-sinhala-0.2.0 which no longer uses surrounding text.
I confirmed this with a local snapshot build of scim-sinhala-trans cvs.
scim-sinhala should be fixed in 0.2.0-1.fc6. Please open a separate bug for si-wijesekera.mim if necessary. (Note that scim-sinhala-wijesekera exists in sinhala cvs on sourceforge which should also solve this problem for wijesekera.)
FWIW I implemented the "surrounding text" callbacks in OOo recently (through a mighty hack), so this had begun to work the same as gedit anyway with the existing sinhala im engine in OOo in rawhide.
(In reply to comment #8) > FWIW I implemented the "surrounding text" callbacks in OOo recently (through a > mighty hack), so this had begun to work the same as gedit anyway with the > existing sinhala im engine in OOo in rawhide. the problem is that it works on gtkimm and probably on qtimm - depends on which version of qtimm we use - but there are no "surrounding text" support on XIM. so it may be not help if one inputs through XIM on OOo, (and OOo is running on KDE if qtimm doesn't support it)
Openoffice it is ok, but for kde, it is not working.
(In reply to comment #10) > Openoffice it is ok, but for kde, it is not working. Which IME is that for? HAve you tested scim-sinhala-0.2.0?
yes, with scim-sinhala: scim-sinhala-0.2.0-1.fc6
Could you attach some screenshots to show an example of the problem? Please include the keyboard input too. You may want to test with scim-bridge-qtimm too to see if that it any better.
> How reproducible: > > > Steps to Reproduce: > 1. open kedit > 2. select scim-sinhala > 3. type ko works fine with scim-sinhala-0.2.0 > 4. with wijesekera > 5. type fdl this has to be fld > 6. repeat step with gedit to see correct output
Tyronne reporetd scim-sinhala-0.2.0 is fine, but not with scim-m17n. Reassigning to scim-m17n. Jens, Mayank, does scim-m17n-sinhala use surrounding text callback?
Hi Leon, The characters "ko" (scim-sinhala) and "fld" (wijesekera) are showing exactly the same ligerature. Do you still need the NEEDINFO? I think this bug is fixed already in Rawhide.
Tested on fc6t3 with following results Kedit - combinations work as expected oowriter - broken rendering - on both kde & gnome desktops. As discussed, lets close this bug for kedit & file a new one for oowriter. Versions tested with oowriter - openoffice.org-writer-2.0.4-3.1 kedit - kdeutils-3.5.4-2.fc6 Pango - pango-1.14.3-1.fc6 libicu - libicu-3.6-1
OO.o issue was cloned to bug 206108.
Created attachment 137130 [details] Problem with wijesekera Scim-sinhala works fine with kedit, while wijesekera does not. qt-3.3.6-13 kdeutils-3.5.4-3.fc6
(In reply to comment #19) > Created an attachment (id=137130) [edit] select M17N-wijesekara open kedit type fld referring to the attached image, when the M17N-wijesekara is used. a dotted circle gets appeared between SINHALA VOWEL SIGN KOMBUWA (U+0DD9) and SINHALA LETTER ALPAPRAANA KAYANNA (U+0D9A). this is a bug, and the circle should get dissappeared when SINHALA LETTER ALPAPRAANA KAYANNA (U+0D9A) is typed. the key sequencec to type in this test case are as follows:fld Pre Conditions : ---------------- Select M17N-wijesekara in SCIM Open Kedit STEP #1 ------- type "f" ; SINHALA VOWEL SIGN KOMBUWA (U+0DD9) appears with a dotted circle. STEP #2 ------- type "l" ; SINHALA LETTER ALPAPRAANA KAYANNA (U+0D9A) appears and the dotted circle, which appeared at STEP #1 should get dissappeared. but this is not happenning. the dotted circle remains the same. that's the bug we have to fix.
As noted in bug 208104 some recent patches to scim-m17n and m17n-lib I believe should allow it to fallback to preedit when the client does not support surrounding-text. I suggest filing a new rfe bug to track qtimm getting surrounding-text support.
Okay, did a small test... Wrote "fld" on a test machine where it was working, scp'd the file to a rawhide machine & then opened it in all 3 - gedit/kedit/oo.o and it shows up perfectly in all 3. Hence I can now confirm that its a scim-sinhala bug. Had it been a font issue, the ligerature had not rendered on any of 3 editors, had it been a pango issue, it must have rendered incorrectly for gedit, same with qt/icu. Jens... anything on this?
Created attachment 137208 [details] test case with fld Open this file in gedit/kedit/oo.o
Oops, i'm completely sorry, I meant its a wijesekera bug (surrounding text maybe).
(In reply to comment #21) > As noted in bug 208104 some recent patches to scim-m17n and m17n-lib > I believe should allow it to fallback to preedit when the client does > not support surrounding-text. Okay, I looked into this a bit further and it is not quite as high level as I had initially optimistically thought. Basically the fallback currently has to be implemented in the map itself: upstream has does this so far in cvs for the Thai maps but not for Wijesekera: so currently either Wijesekera needs to be redone from scratch not to use surrounding-text or better add the the new test for whether surrounding-text is available and provide alternative input processing using preedit when it is not. Ie the new feature is basically just to allow maps to test whether surrounding-text is available or not.
M17N upstream has created a Wijesekera test input map that uses preedit.
Since this is fixed in current release for devel branch, moving to modified.
*** This bug has been marked as a duplicate of 208104 ***
Thanks Jens :)
Mayank, Jens, is the new package m17n-db already included in RHEL5-beta2? Can i now change state of #208104 to Modified since it's duplicate of this bug? thanks
Not really this bug is now intended as a rfe to track adding surrounding text support to Qt IM.
Yes, the builds should be there... as per comment 15 in bug 208104, builds were sent for Beta2 on 2006-11-07.
it's a feature request, so i close it as DEFFERRED.
Ok, but the idea of this bug was to track this for Fedora devel.