Tuesday, July 27, 2010

worked on Meera font today!! (waiting harfbuzz very eagarly)

finally resolved (at least with my testing) Meera Fonts bug https://bugzilla.redhat.com/show_bug.cgi?id=616324

ohh, what a day!!
was trying to resolve this problem
initially thought it will be easier one, just will increase kern for U+0d4d characters, and whenever it will followed by U+200C in will just make it kern to minus.
mine test cases are
1) ക്  (U0d15 U0d4d) should leave proper space, when it will be last character of word (presently chandrakala/virama is getting cut)
2) ക്‌ക (U0d15 U0d4d U200c U0d15) when virama/chadrakala will be followed by 200c it should use present kern to get form in word

so tried first dist feature first,
1) increased kern of U+0d4d and
2) added pair positioning for "U+0d4d  ZWNJ" with dist flag kerning
was working well when i was debugging in fonts but failed when was testing with pango :(

next try was
then added ligature and was trying to bring that ligature when someone type
Cons + 0d4d + 200C
but again it failed in pango :(
tested same in Open Office and it was working there fine, with some more trial understood that pango is not considering 200c for contextual comparison
so writing rule i.e when U+0d4c is preceded by ZWNJ will not work in pango

next try
since it was need to increase kern of virama only when it will be last character of word,
and luckily while trying contextual feature found {Everything Else} group, so used this one
and it worked perfectly in pango :)
but again problem, it was not working in oowriter

so added both 2nd and third solution in font for making it work with oowriter and gnome
link for
1) scratch build is available on http://koji.fedoraproject.org/koji/taskinfo?taskID=2353995
2) modified source file is @ http://pravins.fedorapeople.org/Meera.sfd
3) binary file for testing @ http://pravins.fedorapeople.org/Meera.ttf
4) patch: http://pravins.fedorapeople.org/bug-616324.patch   

dont know how dream of harfbuzz will come true, and will make font engineers life easy

Wednesday, July 21, 2010

New Released Liberation Fonts!!

Just done 1.06.0 release of Liberation Fonts.

Highlights are:
1) New family added to Liberation-Narrow Fonts (Thanks to Herbert Duerr for contributions)

2) Some changes from releasing side

Now releasing source tar ball as well, so if someone want to build from source can use this tar ball.

If someone want to direct use ttf file, he can download Binary(ttf) tarball, and use it directly.

Added TODO list as well for next release :)

1) resolving bug related with hinting
    - https://bugzilla.redhat.com/show_bug.cgi?id=606217
    - https://bugzilla.redhat.com/show_bug.cgi?id=591556
2) shape improvement
    - https://bugzilla.redhat.com/show_bug.cgi?id=591559
    - https://bugzilla.redhat.com/show_bug.cgi?id=487581
3) Ascent Descent Values Improvement
    - using absolute values instead of relative values in OS/2 table
4) RFE: Add Greek Polytonic support to Liberation fonts
    - need some volunteer to add these shapes
    - https://bugzilla.redhat.com/show_bug.cgi?id=473842

 really reproducing these hinting bugs is big task for me

Will build it for fedora tomorrow

Monday, July 19, 2010

Added New Indian Rupee symbol — INR to Lohit Devanagari

Friday saw the news on new symbol for INR, i felt really happy since the prevoious one U+20A8 ₨ was based on latin shortname for Rupees, and never it was looking like symbol. Just capital R and small S combination.

That's why it was really less in use, and many people were not aware of this. While checking in locale i found this one only used in en_IN locale. Other locale were using locale initial for rupees. i.e in Marathi and Hindi रु. was used for currency symbol. Same for Other language in there own script.

Standardization is really favourite topic of mine and really with this one we can standardize INR symbol.

Now all developers are waiting to get Unique Code Point for this symbol. i.e Unicode Value. We have for all other currency symbols see http://unicode.org/charts/PDF/U20A0.pdf

I am strongly recommend to add this symbol also on same Unicode code page.

in excitement /me also added this characters in lohit devanagari fonts and committed to upstream svn :)

Since no Unicode values is assigned to this, there is no standard for typing or storing this character.

I just added so if any person want to give reference to this symbol can easily use lohit fonts and do so

I have added this character on U+E000, which is unicode private user area.

So anyone want to type and use this characters just enter U+E000 with any method you know or ask me here :)

This is private use area so we can use this for any general purpose, please note when we will get actual value we just need to replace U+E000 with the new unicode value, thats it

I saw some people http://blog.foradian.com/ added new INR symbol on ` location, but this is one kind of unicode spoofing.

Since in actual storage you are going to store ` (U+0060) and whenever you want to type ` character you will only get new INR symbol.

So this is wrong thing and i recommend to use some private user area location, so it will not conflict with existing Unicode Encoding Standard

Download Lohit Font with addition of new INR symbol at U+E000

for more information about Lohit project see  see https://fedorahosted.org/lohit/