tiro bangla font showing wrong display of some glyphs

[TeerthoMukhopadhyay]

In tiro bengali font, it is shoing non-ended vertical line on some glyphs such as ঋ ৠ খ গ ঝ ণ থ প শ গ্রু গ্রূ থ্রু থ্রূ প্তু প্রু প্রূ শ্রু শ্রূ খ্ব খ্র গ্গ গ্ন গ্ব গ্র গ্ল ঝ্র ণ্ণ ণ্ব ণ্র ত্থ থ্র ধ্ন ধ্র প্ত প্ত্র প্ন প্প প্র প্ল ব্ঝ ম্প ম্প্র শ্ত শ্ন শ্ব শ্র শ্ল ষ্প ষ্প্র স্খ স্প স্প্র স্প্ল ড়্গ শ্ত শ্ন শ্ব. Also showing unnatural conjunct design such as ন্তু ন্ক্ত ন্ত ন্ত্ব ন্ত্র ন্থ ন্থ্র. Also showing absence of ZWJ consonant forms such as ক্‍ খ্‍ গ্‍ ঘ্‍ ঙ্‍ চ্‍ ছ্‍ জ্‍ ঝ্‍ ঞ্‍ ট্‍ ঠ্‍ ড্‍ ঢ্‍ ণ্‍ থ্‍ দ্‍ ধ্‍ ন্‍ প্‍ ফ্‍ ব্‍ ভ্‍ ম্‍ য্‍ র্‍ ল্‍ শ্‍ ষ্‍ স্‍ হ্‍ ড়্‍ ঢ়্‍ য়্‍ ৰ্‍ ৱ্‍. Please reform that glyphs. A screenshot is here--

After all, correct this glyphs. You can get idea about the concept in Noto Serif Bengali font. Thank you.

[tiroj]

1. The short vertical spur is a characteristic of some styles of Bangla lettering. Tiro Bangla is a different style that does not include this feature. This is by design.

2. The shortened half-form-like glyphs that occur for some letters in some Bangla fonts are not a traditional feature of the script. Half-forms are an idea imported from simplified Devanagari typesetting, and do not apply well to the more complex models of Bangla conjunct formation.

[TeerthoMukhopadhyay]

This fonts are designed as a book font, right? the vertical spur of book fonts should not be shortened, there are many many fonts as reference. If this is a "design", it should not be able to publish as a book font typeface like this. And devanagari writing style is not so similar to bengali. Bengali has some extraordinary halfform characters. in the font, the half form of ষ, স, and more have a changed style. So ন-halfform should also have that changed style, not as devanagari. As a book font, people deserved it as a real south asian script font, so the realness should achieve.

Actually, my request is--

1. The font style is book font style and every bengali book font should have the expanded vertical spur in glyphs also for tiro bangla. it is not a stylish font, it is a book font.
2. the half form of ন should generated as bengali traditional style because another half forms like ষ, স, etc are generated in traditional style.
3. the zwj forms of bengali consonants should add for calling academic font conjunct style without adding extra conjunct glyphs.
4. the word final form should not apply in only া, ী, ো, ৌ but also in ঋ খ গ ঘ ঝ ণ থ ধ প শ for keeping the form accurate.

After all, this edits will make the font more good and comfortable with right style.

[tiroj]

This typeface has been used by many people who like it, and it has been used in prestigious book publications. If you don’t like the style, you don’t have to use it. There is plenty of scope within Bangla script for a variety of styles, and for different ways of handling conjuncts. Again, we deliberately avoid use of half-form glyphs because they are not a traditional feature of the script, but are a technical mechanism borrowed in some fonts from Devanagari typesetting technologies.

[tiroj]

That said, we’ll discuss this feedback internally, particularly regarding the ন্ত ন্ত্ব ন্ত্র ন্থ ন্থ্র conjuncts forms.

[tiroj]

4. the word final form should not apply in only া, ী, ো, ৌ but also in ঋ খ গ ঘ ঝ ণ থ ধ প শ for keeping the form accurate.

I will definitely consider this. The final forms had to be disabled in the original Murty version of the typeface, and then only applicable selectively using a stylist set feature, because of issues in Indic shaping engines not applying contextual lookups across cluster boundaries. This is still an issue with Microsoft’s DWrite engine, although fixed in HarfBuzz and in Apple’s CoreText engine. So while final form substitutions now work reliably in many places, they don’t work everywhere, and I am cautious about adding more while this is the case.

[TeerthoMukhopadhyay]

Please try to understand. Many people shared this unincompatibility with me and i also being a bit frustrated with this worse issues.

1. The shortened vertical spur is looking poor with া which has full vertical spur. it should be shortened or shortened vertical spurs should be expanded. i suggest the full vertical spur form for consonants. i have discussed with many bangladeshis and they also suggested it.
2. the ন্ত ন্ত্ব ন্ত্র ন্থ ন্থ্র conjuncts forms should reform because the half form of ন is looking like half form of প or শ but the half form of ন should look like half form of ষ or স, the vertical spur of half form of ন should be rounded like noto serif bengali. the half form of ষ and স uses traditional bengali styles. so half form of ন should also use the traditional rounded vertical spur style not like academic linear vertical spur style.
3. the zwj form of consonants should not avoid because some of the bangalee (like 30%) people want to write in clear academic conjunct form. when there is not academic conjunct form basically, then zwj forms of consonants should produce for academic conjunct forms. the noto bengali fonts also produced that. it should appear for completion of glyph chart of tiro bangla. Bengali also have the zwj forms but not in all fonts which are called like poor fonts.
4. the word final form cannot be seen in windows 7, i tried but failed and don't know why! word final form reproducing is an attractive feature but not applyable in all devices, right? then try to make the word final form for applyable devices. some devices are not showing word final form but it is temporary but good solution to produce wordfinal form. Some device will accept that and some cannot, nothing to do (!).

[tiroj]

Thank you for your patience. I have been thinking through some of these issues, and will discuss with my colleague and co-designer Fiona Ross after the holidays.

The good news is that, since this is an open source project, there is the possibility of forking a version that conforms to your feedback even if we’re not able to implement all the suggestions in the main branch.

1. This is something I will discuss with Fiona. The vertical spurs are not something we simply overlooked, but were a deliberate design decision create less fussy, cleaner shapes and also to make a distinctive design that wouldn’t simply reproduce shapes and conventions from Fiona’s earlier Linotype Bengali fonts. We wanted to make something new, slightly less traditional, with less going on above the head-line stoke.

2. This is something I will consider. One constraint is that we can’t make any changes in the main branch fonts that would affect advance widths and spacing, because that would cause text reflow in existing documents. So any change in the ন component* shape would need to fit within the area occupied by the current shape. I note, by the way, that in some typefaces the shape of this component is irregular, e.g. Linotype Bengali

3. I am aware that use of half-forms in Bengali is increasing, and that they’re being taught as such in some contexts. I remain unconvinced that they’re a good idea, since the nature of Bangla conjunct formation is seldom as linear as it is in Devanagari, and even that minority of conjuncts in which the component parts might by analysable as half+full forms often involve small adjustments to proportions and the way the components join. Also, we use contextual substitions and positioning to apply appropriate ikar width variants and to adjust the position of chandrabindu and other marks, which can be done more precisely on a ligature glyph than on a decomposed half+full form sequence.

If there are conjuncts attested in Bangla (or Assamese, Manipuri, etc.) language texts that are not supported as ligatures already, I would really like to document them. That is true regardless of how we end up implementing them at the font level.**

4. I documented this problem with support for cross-cluster contextual rules in Indic shaping engines back in 2013 [Problems for Indic typography in current OpenType Layout implementations], and the following year I was able to get engineers from Microsoft (DWrite), Google (HarfBuzz), Adobe (Lipika), and Apple (CoreText) together to discuss the issue and come up with a solution. That solution was published in 2014 [Fixing Indic2 OpenType Layout], but it was only implemented in HarfBuzz and CoreText. Microsoft didn’t implement it in DWrite, and Adobe failed to implement it in Lipika, but in recent versions of their CC apps they have switched to using HarfBuzz for Indic scripts, so now Microsoft is the only company that still has broken support for word final forms or other typographic features that rely on cross-cluster conextual rules.

---

• I am deliberately avoiding use of the term ‘half-form’ in this context, because it is misleading. I consider half-forms to be a specific technical implementation of typesetting of some Indic scripts (notably Devanagari), and not a useful term in discussing either writing or other forms of conjunct composition. In the case of Bangla, we implement conjuncts using cjct feature ligatures, without any half-form glyphs being involved.

** One day, I would very much like to make a Bangla font that is capable of handling Sanskrit texts written in the script, but I am aware from my research on Sanskrit conjuncts that this will involve many forms for which no examples exist in Bangla script.

[TeerthoMukhopadhyay]

Okay, i am so happy that you have considered my request.

1. Actually the vertical spur of consonants is seeing unnatural with া and i want the same rule both for that consonants and aa-kar.
   
   See, the vertical spur rule should show like the font sarkar or like the font noto serif bengali for comfort. the full vertical spurs are more comfortable to read. Creating something new should not be like changing rules of bengali characters. Theoritical anatomy is here--
   
2. the half form of ন should not have a linear vertical spur, it should have the curved spur. See this--
   
   the half form of ন is curved in other fonts except this tiro bengali font.
3. the half form of consonants are available in other good fonts. Please Consider to produce half forms in tiro bengali font. Every fonts are using the feature so why tiro bengali not? half forms (ZWJ forms) of consonants will help to make more clear conjuncts manually.
   
4. But can you produce the word final forms? devices which do not allow the form will not allow for any vowelsigns but word final form of vowel signs were made! so word final form of consonants did which sin?
5. Hey, there is regular and italic version of the font, but why there are not the bold and bold italic version??? bold version is needed for document publication cases.

[tiroj]

re. 4: The issue with the word-final forms is that without correct shaping engine support they may be triggered within a word, because each cluster is processed as an atomic entity, so there is no distinction between the end of a word and the end of a cluster.

re. 5: the Murty Library publications for which the fonts were originally developed did not use bold weights, so these were not commissioned. Bold weights will be available in future, but probably only under a commercial license unless we can find a source of funding to release them under the OFL.

[TeerthoMukhopadhyay]

what about issue 1, 2, 3? about issue 4, i want to say that please experimentally publish the font with word-final forms and try to apply that perfectly in font engines. the independent kerning vowel signs are going rightly in android. Also make the word-final forms for other glyphs in the same way. But publish that experimentally. about issue 5, i want to say that the bold version of tiro bangla is actually needed for book publications but it is absent, in the way it will good to publish bold weight under OFL license. Most of the paid fonts or font weights are hacked by bad people. So there is no way to publish the bold weight paidly. Dear John Hudson, Please take it with full patience. Best of Luck.

[tiroj]

As I said previously, we’re going to consider the other issues. I expect to report back in a few weeks, but it will be some months before we would be able to make any updates to the fonts, as we have a lot of other work scheduled.

about issue 4, i want to say that please experimentally publish the font with word-final forms and try to apply that perfectly in font engines.

This cannot be done in a way that will work with Microsoft’s DWrite engine, because cluster segmentation is outside the role of the font. Unless the Microsoft engine is updated to behave the same way as HarfBuzz and CoreText, we can’t automatically apply word-final forms.

To explain in more detail: no word-level analysis is being performed by any engines to identify the end of words as apply these forms. So in order to apply them via the font lookups, we have to use a negative context rule, i.e. ‘Substitute these forms when not followed by any other letter glyph’. But this produces incorrect results if an engine does not apply contexts across cluster boundaries—as is the case with the MS engine—, because each individual cluster terminates as if there is no followng letter glyph. This produces final forms within words, with unwanted gaps in gthe head-line stroke.

[TeerthoMukhopadhyay]

i want to say that if the word final forms of aa-kar, ii-kar and other kerning vowel signs were made but cannot be applied in ms dwrite engine, then in the same way, for other glyphses word final form can be made. the word final form of kerning vowel sign are not working in ms dwrite enginee, i have tested that. so in the same way, the word final form can be appied for other glyphs and the form will not apear in microsoft dwite enginee. After all, the word final forms can be made in the same way of kerning vowel signs, and microsoft dwrite engienee will show the full forms all time, as shown in the case if kerning vowel signs.

[tiroj]

The issue in apps like MS Word is not that the final forms are not displaying, but that they are displaying where there are not wanted, e.g.

I am wary about spending time adding a lot of additional final forms while this remains a problem, because it would break up every word into into lots of separated clusters. I am stll not sure that it was the right idea to implement the final forms for the vowel signs in this situation, but did so because two out of three of the major shaping engines supported it, and we don’t get progress without pushing.

[sridatta1]

the half form of consonants are available in other good fonts. Please Consider to produce half forms in tiro bengali font. Every fonts are using the feature so why tiro bengali not? half forms (ZWJ forms) of consonants will help to make more clear conjuncts manually.

Noting the half forms in Noto fonts are removed in the latest releases. though it's unrelated to Indigo project.
I have also raised a ticket on the same, to not to support them.
notofonts/bengali#10
As Indigo is an open-source is an open source project, you are free to fork and modify the designs the way you want

[TeerthoMukhopadhyay]

@sridatta1 why kicking zwj forms going valid? some people uses zwj forms for making more clearer conjuncts. it is valid to make zwj forms-- the work of zwj is this. in bengali keyboard layouts, the zwnj and zwj key are available. so the proper work of zwj should be formed. dont reject the support of zwj forms, it is the normal behaviour of bengali fonts.

Noto fonts also should have half forms. dont kick them because it is valid and not unrelated.

@tiroj, as the general mindlaw, i accept your description. then, can you make wordfinal forms for those enginees who supports the forms correctly (except for ms word)? and also i am feeling the necesity of bold weight of the font. I want to say that please make the bold weight of the font and make them free, not paid font, if you can.