Description of problem: In case of conjuncts with Yaphala (U09CD+U09AF), multiple Yaphala signs can be typed consecutively giving rise to invalid conjunct combinations. e.g ক্য্য্য্য (ক+্+য+্+য+্+য+্+য) This error could be replicated in gedit, kwrite and Openoffice.org Writer Version-Release number of selected component (if applicable): fonts-indic: 2.3.8-1.fc11 gedit : 2.26.1-1.fc11 openoffice.org: 3.1.0-9.2.fc11 kdebase : 4.2.2-2.fc11 How reproducible: Steps to Reproduce: 1. Open gedit 2. Type in the following unicode values (for the above example) : U0995+U09CD+U09AF+U09CD+U09AF+U09CD+U09AF+U09CD+U09AF Actual results: The last 3 U09CD+U09AF pairs are allowed as part of the conjunct Expected results: The U09CD+U09AF pair should not be allowed to be rendered after another U09CD+U09AF. i.e. only one U09CD+U09AF pair can be used when combining with another consonant to form a conjunct. Additional info: Please do note that: U09AF+U09CD+U09AF is a valid conjunct combination, where two U09AF characters combine to form a conjunct (য্য) Screenshots: gedit: http://runab.fedorapeople.org/assorted-bugs/gedit-ya.png kwrite: http://runab.fedorapeople.org/assorted-bugs/kwrite-ya.png oowrite: http://runab.fedorapeople.org/assorted-bugs/oowrite-ya.png
IMO, this is not wrong semantically, since a syllable can be composed of any number of Consonant+(Halant+Consonant)n sequences. Although this can be a case of not usable syllables or the syllables that are not used in the given language, it may not qualify as a critical issue to be tweaked in the layout engine. @Amit, can you please mention if this is a bug for Assamese language as well or just an unused syllable?
This bug appears to have been reported against 'rawhide' during the Fedora 11 development cycle. Changing version to '11'. More information and reason for this action is here: http://fedoraproject.org/wiki/BugZappers/HouseKeeping
me too agree with Rahuls comment can we close this bug?
after discussion with runa on this bug found that this issue can be solve in font itself by just adding some contextual GSUB rules, context will be before executing rule for halant+ya check does it preceded by halant+ya if yes do not execute rule else execute it but bit doubtful about do Open Type provides such a tag for bengali script like we have for Arabic contextual http://www.microsoft.com/typography/otfntdev/bengalot/features.aspx i will back to this
Submitting a patch to this bug. Please verify it: Patch: http://nkumar.fedorapeople.org/lohit-bengali-fonts/Lohit-Bengali-Bug497090-YaphalaConjuncts/Lohit-Bengali-Bug497090-YaphalaConjuncts.patch Modified sfd: http://nkumar.fedorapeople.org/lohit-bengali-fonts/Lohit-Bengali-Bug497090-YaphalaConjuncts/Lohit-Bengali.sfd
Created attachment 398759 [details] Demonstrates what this patch solves in Lohit-Bengali-Bug497090-YaphalaConjuncts
Created attachment 398760 [details] Patch to this Bug
TTF file: http://nkumar.fedorapeople.org/lohit-bengali-fonts/Lohit-Bengali-Bug497090-YaphalaConjuncts/Lohit-Bengali.ttf
Hi Naveen, Thanks for the patch. On gedit the display is correct as per the expected result. However, for kwrite and OpenOffice.org Writer, boxes are displayed instead of a dotted circle. Images of the actual displayed characters are as follows: kwrite: http://runab.fedorapeople.org/assorted-bugs/kwrite-497090.png oowrite: http://runab.fedorapeople.org/assorted-bugs/oowrite-497090.png Thanks Runa
h(In reply to comment #10) > Hi Naveen, > > Thanks for the patch. On gedit the display is correct as per the expected > result. However, for kwrite and OpenOffice.org Writer, boxes are displayed > instead of a dotted circle. Images of the actual displayed characters are as > follows: > > kwrite: http://runab.fedorapeople.org/assorted-bugs/kwrite-497090.png > oowrite: http://runab.fedorapeople.org/assorted-bugs/oowrite-497090.png > > Thanks > Runa Thanks for replying. Here is the quick fix to that Kwrite and oowrite box problem. It is fixed for Pango but it looks like it is difficult to fix it for Kwrite & oowrite (icu). Also I have restored the problem for Kwrite & oowrite (icu). Now no more boxes but it is not patched for Kwrite & oowrite (icu). They don't seem to read Contextual Substitution Tables well or not at all. Patch URL: http://nkumar.fedorapeople.org/lohit-bengali-fonts/Lohit-Bengali-Bug497090-YaphalaConjuncts-1/Lohit-Bengali-Bug497090-YaphalaConjuncts1.patch SFD URL: http://nkumar.fedorapeople.org/lohit-bengali-fonts/Lohit-Bengali-Bug497090-YaphalaConjuncts-1/Lohit-Bengali.sfd TTF URL: http://nkumar.fedorapeople.org/lohit-bengali-fonts/Lohit-Bengali-Bug497090-YaphalaConjuncts-1/Lohit-Bengali.ttf I tested it on a Windows Machine and it looks exactly like pango output (tested it on pango-view & gedit). Need some more time to check if it could be fixed at all for Kwrite & oowrite (icu) from font side. Otherwise need to raise a bug for QT and ICU.
This bug appears to have been reported against 'rawhide' during the Fedora 13 development cycle. Changing version to '13'. More information and reason for this action is here: http://fedoraproject.org/wiki/BugZappers/HouseKeeping
I had another solution to this using class based contextual rule, which could have worked even for QT/KDE applications, but then @Runa showed me that it introduces ZWNJ/ZWJ problems. And Pragmatically, I personally believe, bugs like these are less important for they are more about invalid sequences than a valid one. I am expressing this view only by taking into account failed attempts by us (indic developers) to push changes/patches into rendering engines. The situation is grim, but let's face it.
yes, these combination doesn't occur in really data (i.e requested in expected results) and as tried by Naveen and Runa, resolving this introducing regression. thanks naveen for trying this, surely it improved your knowledge of OTF Contextual GUSB:) closing this