bugFree UCS Outline Fonts - Bugs: bug #51810, Bengla: phala form problems

 
 

bug #51810: Bengla: phala form problems

Submitter:  Steve White <Stevan_White>
Submitted:  Mon 21 Aug 2017 10:56:39 AM UTC
   
 
Category:  individual character(s) Severity:  3 - Normal
Item Group:  character substitution issue Status:  Fix posted
Privacy:  Public Assigned to:  Stevan_White
Open/Closed:  Open Release:  * development
* Mandatory Fields

Add a New Comment Rich Markup
   

Mon 04 Sep 2017 04:37:48 PM UTC, comment #3: 

Implemented some ideas in Sans, but it's complicated.
Formed, ra, la, ba, na and ta as 'blwf' "pholaa"s, just because they usually went under other consonants in combinations.
Process to form "ligature" glyphs is complicated:

Usual horizontal form: "half" + normal.

Vertical: for non-pholaas, have to make last consonant a below-form.
For ones with a pholaa, have to make the preceding consonant a "top" form (reduced but preserving vertical stem).  For ones without pholas, make the second consonant a "below" form.
And then there are exceptions, like "ba-ba", where it is preferred that the second consonant be full size.

Steve White <Stevan_White>
Group administrator
Mon 04 Sep 2017 09:15:39 AM UTC, comment #2: 

For the record: about phala forms.

The word phala or falaa means "point", primarily, an arrow-tip.  It is also used in Bengali writing to mean an annex to a letter.  The Unicode standard and the OpenType documents call the special forms of ra and ya phala, but the OpenType standard recommends using the "blwf" table also for the below form of ba.

So, which letters exactly have phala forms?  I don't think the answer is simple or easy.

Halhed's 1778 Grammar of the Bengali language lists 12 "P,holaa"s.  It's important to know that, at least at that time and to the best of Halhed's knowledge, the notion of "P,holaa" was quite broad.

He lists:
   ya ra la na ma
Note that in his book, ma takes a wholly subscript form, while more recent forms have a right vertical to the right of the preceding letter.
He spends a lot of time fussing with the "ba", which is conflated with the Sanskrit "wa" or "va", and earlier Bengali versions of these consonants.  This remains a Unicode encoding lacuna at this time.  Anyway, he regards these also P,holaa.
He also lists the subscript vocalic r -- well, it's also a below-form.
But he goes on to list left-combinations with sha and nga -- which don't produce below forms now or then...  They seem to be a sort of instructional device, referring to sets of consonant combinations, rather than a simple set of subscript glyphs however.

How does this impact the practical implementation of the font?  This also isn't clear. The ya has a lookup specially for it: 'pstf' -- but the modern ma and also kha too behave in a similar way -- sometimes.  The OpenType docs show the 'blwf' lookups being used for ba and ra phala forms -- but it might be advantageous to handle other consonants here too.  However, each entry here has to be taken into account in later tables: since 'blwf' executes befor 'half', it absorbs the intervening halanta so the 'half' form of the preceding consonant does not form.

In any case, for 'bng2', it's essential that every consonant of every conjunct be processed through one of 'blwf' 'half' or 'cjct' in order for the conjuct to be recognized as such, and for a closing i-vowel sign to be placed properly.





Steve White <Stevan_White>
Group administrator
Tue 29 Aug 2017 10:54:49 AM UTC, comment #1: 

There are several cases of conjunct ligatures that, according to the TDIL document, ought to be vertical, that are rendered in horizontal form, and vice-versa.

Note the TDIL document may not be perfect.  In particular it displays two conjuncts, ঘ্ঘ and ঢ্ঢ, with hasanta.  Indic language speakers have told me these conjuncts are very rare, if used at all.  Different fonts display them sometimes with and sometimes without a hasanta.  I guess this was an oversight in the TDIL document, and will arrange for them to be displayed horizontally without hasanta.

Steve White <Stevan_White>
Group administrator
Mon 21 Aug 2017 10:56:39 AM UTC, original submission:  

Some phala forms look funny in sans, for example, dd-ma, the ma seems to in phala form, but crashes into the dd.  There is a chain lookup that is supposed to sort out the proper phala form, but it's not working optimally.

When should a conjunct take a phala form at all?  One sees mention of phala forms for ra- ya- ba- and ma-, but also la- clearly forms something like a phala.  (Ra is sometimes written 'rofola')

According to the TDIL Bengla Grammar, most conjuncts don't occur in Bengla: it says they should be rendered in "linear form", by which I guess it means with a half-form first and either regular or phala form second consonant. 

Besides standard modern conjunct ligatures, the TDIL document notes "traditional" ligatures.  The font does not make this distinction.

Some other fonts skip the phala forms for most leading consonants.  We took the position that, even if the conjunct was not used, it should look "right" even if mis-typed, and that meant that phala forms should appear.

I think when I made this, I had limited resources -- I don't think I had the TDIL document.  A PDF using the font was perused by native speakers -- but perhaps it didn't contain the problem conjuncts, or maybe they weren't sensitive to the funny phalas.

Steve White <Stevan_White>
Group administrator

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

Attach Files:
   
   
Comment:
   

No files currently attached

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -email is unavailable- added by Stevan_White (Submitted the item)
  •  

    There are 0 votes so far. Votes easily highlight which items people would like to see resolved in priority, independently of the priority of the item set by tracker managers.

    Only logged-in users can vote.

     

    Follow 2 latest changes.

    Date Changed by Updated Field Previous Value => Replaced by
    2017-09-04 Stevan_White StatusProceeding Fix posted
    2017-08-29 Stevan_White StatusNeed info Proceeding

    Back to the top

    Powered by Savane 3.13-d3ae.
    Corresponding source code