Thursday, September 19, 2013

Before the alpha release of Lohit Devanagari from Lohit2 project

   I am hoping now most of the contributors around are aware regarding the lohit2 project. Before the Alpha release of Lohit Devanagari i think it is important to go through once again goals we planned for this project[1].


    1. Cleaning Lohit Open type tables.
    Highlights are as below
  • We rewritten all GSUB rules from scratch.
  • New rules are supporting both deva and dev2 script tag
  • Done testing on Harfbuzz as well with Uniscribe and its giving expected results.
  • Kept GPOS tables intact.
  • Effectiveness and efficiency [2]  (sfd file size is down by 28K and Binary file side is down by 4K)
    Found one bug w.r.t harfbuzz [3] and looking forward to get it resolved soon. Presently it is in know issue list.

    By Beta we will have some more improvement on this.

    2. Reusable Open type tables.
    We got two important suggestions on this line, so below are suggestions and action taken on it.
    1st suggestion
    To have feature file separate than shapes .sfd file for easy re-usability of OT rules.
  • Thanks to AravindaK, he has already done some work on that line[4], so just using those stuff. I have forked this gitrepo and doing some improvement in it. Once done will request Aravinda to merge with his repo.

    2nd suggestion
    To follow AGL[5] and to have readable glyph naming.  We were also thinking from this perspective.
  •   This has became a bit complex glyphlist.txt [6] suggest names like "kadeva" or uni0915. But we dont want to follow uni0915 as it is not very readable considering our re-usability goals.
  • {0915 (kadeva) + 094D (viramadeva) + 0937 (ssadeva)} following this create chances of glyph name string more than 31 characters limit.
  • So present plan is follow above "kadeva_viramadeva_ssadeva" as much as possible and if it goes above 31 characters we will discard "deva" part from glyphname.

    3. Following of existing standards/guidelines
    Dont know how many of you aware regarding "Devanagari Script Behaviour For Hindi"[7] Draft, so its basically guideline for Font developers. I have one blog pending on this. Though this is draft mode we are trying to follow this, since it is very informative and prepared after consulting to language experts.

    This is where we upto, if anything more needed do provide me your feedback. Also need to decide on release version, i think some version with -alpha will work.

2. Effective means it should work on all supported platform perfectly and efficient means compact and clear rule


R. said...

A suggestion I have on '2nd suggestion' is to name glyphs with script tag in the beginning and unicode name - deva_ka.
And name the complex glyph deva_kassa, omitting virama in the naming.


Pravin Satpute said...

deva_ka thing is doable but unfortunately second suggestion by you about deva_kassa is not as per Adobe glyph naming guidelines.

We are following kadeva since it is given by AGL

We had nice discussion on AGL and the concept is very clear to me now. Might be you would like to go through

I will soon write one blog specifically for AGL.

Tutin Corin said...

जैसे मराठी पेपर लोकसत्ता का "ल" आता है, वैसा हिन्दी का कोई फॉन्ट कोई है क्या? यूनीकोड में कोई विविधता ही नहीं है, इस बारे में कोई कुछ काम कर रहा है क्या? कृपया बताने का कष्ट करें. हिन्दी का पूर्नविराम भी नहीं मिलता है!!!

Pravin Satpute said...

मराठी "ल" और हिन्दी "ल" सिर्फ glyph variants है ।

लोहित देवनागरी फॉन्ट मै दोनो glyph variants मौजुद है।

और आपके ऑपरेटिंग सिस्टिम के भाशानुसार "ल" variant आ जाता है !!

uni+0932 for this purpose one can refer to lohit devanagari fonts. We are already handling these stuff from long time with "locl" tag of open type. there are some other glyphs like uni+0936 श. It also has different shape as per language being used.