Handling variations in forms of Devanagari conjuncts.

[mehulchanpura]

[Removed original bug report related to contextual ikar width selection interaction with stylistic set conjunct variants in Hindi and Marathi fonts. See original Issue #63 for details and fix report.]

[mehulchanpura]

Thank you very much tiroj for providing me dev version 1.62 of this font.

I am not a font creator, I am just trying to modify existing font as per my requirement. I am very new in this field so as long as I am working I have some basic types questions like, if I create Substitute of any glyph, suppose if I create a substitute glyph like dTtPa.ss02 for dTtPa for to replace it when i will typing, now with your dev font file I can easily work on this, but recently I notice and confused that dUdatta with dTtPa is placed above on last like dPa.

I will modify font in different names so I can share a screenshot of typed text where I get confused that which one is correct.

Honestly I don't know if this is a right platform to ask (know) how this things should be works.

I am sharing screenshot of four different types of combinations. First one is as default, and then next are different types of substitutes with Udatta plus dSignI.

I am really very sorry if i ask my confusion here in this platform, but I want to learn and contribute how can I or we can modify native font as per our requirements.

It will be pleasure for me if I get knowledge how dUdatta must need it's position if we create a substitute of any glyph in devanagari, I just want to make sure that is dUdatta is strictly followed by his native character or we can use or put it on the last character of the combination?

I am sorry again if I ask this question in this platform and heartily sorry if I waste developers' time.

Now here is the screenshot which I mentioned. Please guide me.

[tiroj]

My understanding is that the udatta mark is normally input at the end of a conjunct (and following a vowel sign if present). As is typically the case with stress accent marking, the sign applies to the produnciation of the consonant+vowel[+consonant] syllable, hence it always comes at the end of the cluster. This is an expectation of Unicode text processing, and an udatta input before a vowel sign will be flagged as an invalid sequence by shaping engines:

So I build my fonts with that assumption, and include contextual positioning rules to move the udatta slightly right and/or up to clear the vowel sign, as you see here

[This is Tiro Devanagari Hindi, in the Sanskrit fonts the mark is positioned differently, above vowel marks, presuming Vedic accentuation.]

I am not quite sure how you are inputting the sequences that result in these shapings, but neither looks right:

Devanagari shaping engines follow a modern Hindi rule in which the ikar will not reorder past an explicit halant, so the if I user ZWNJ (U+200C) to force the presence of an explicit halant, what I will get is

(with the udatta still input at the end of the cluster, after the vowel sign).

[mehulchanpura]

Thank you very much tiroj for you detailed answer.

I think I may made my question much complicated and without add details how I did input sequences. My English is not that good but I am still trying to understand and asking my question in English, so I can get solution of my doubts.

Let me re explain my question again, your font already have dTtPa glyph, but while I am working on transcription or using OCR for to convert and type those devanagari texts in unicode, many times I need to type dTta+zero widhth non joiner+dPa, so I thought what if I will modify font and create a substitute of dTtPa as dTtPa.ss02, and I made this and looks like this in my modified font.

Then I thought what if this dTtPa.ss02 have with dSignI matra (i mean like dTtPa.ss02+dSignI), so after creating dTtPa.ss02 I add this glyph in 'pres' Pre Base Substitutions in Devanagari2 lookup 11 contextual 2, and after that if I type dTtPa.ss02+dSignI, it's look like this,

Then when I am checking positioning of Anchor points, I notice there is a anchor for dUdatta, and here I get confused, that where should I need to position dUdatta in this case, at the top of the last on Pa, and that's why I positioned dUdatta like this,

If I will not create dTtPa.ss02 visually like above screenshot, in that case I need to to type dTta+dHalant+zero width non joiner+dPa+dSignI+dUdatta, and in this sequence it will be visually look like this,

In this case I can type this sequence without modify your font, but there are some other glyph I must need to create a substitute for them to visually look same like my pdf while I am typing in MS Word. Normally I still didn't need requirements to type like dTta+dHalant+zero width non joiner+dPa+dSignI+dUdatta, but there is question in my mind what if I will need this sequences in future! and that's why I am confused.

After reading your reply, I am not that good in English to understand, but as long as I understand and then one more question rise in my mind is that not possible this type of sequence in modern devanagari, like dTta+dSignI+dHalant with or without following dPa or any other devanagari character? is it really not ideal to type dTta+dSignI+dHalant, I mean this not possible in modern Devanagari?

It's true that I am asking because lack of my knowledge about how to write Devanagari/Hindi language. In Unicode font there are anchors in glyph and those all anchors are following their assigned sign, so I got confused what if these signs will I need to use with any of the substitute glyph added by me in my modified font or with combination using zero width non joiner or in future what if I will need to use any of these anchors assigned sign with my modified .ss02 substitutes. This is my concern and I have no mentor, I am just trying to learning on base of trial & error.

I know without mentor or without any faculty guidance it's almost impossible to understand whole concept, but still I am trying and thank you very much for your detailed and kind reply.

[tiroj]

Then when I am checking positioning of Anchor points, I notice there is a anchor for dUdatta, and here I get confused, that where should I need to position dUdatta in this case, at the top of the last on Pa, and that's why I positioned dUdatta like this,

My understanding is that this is correct if one is displaying the conjunct sequence with explicit halant.

Regarding ordering, the standard order for Devanagari clusters following the Unicode model adapted from ISCII, is

Consonant [nukta] [halant] [Consonant] ... | Vowel Sign | Accent

This is a kind of phonetic ordering, following the order in which the characters are pronounced rather than the order in which they are written. Some sequences that do not follow this standard order may be marked as invalid by a shaping engine, e.g. applying a vowel sign directly following a halant, but some others may be permitted because of odditites in some orthographies. So the sequence you describe in which the halant follows the ikar can be cleanly rendered टि्प but this is problematic because it reads the same way as ट्पि, but now there is confusion about how the sequence is encoded. I recommend following standard encoding order practice.

[mehulchanpura]

Thank you so much tiroj.

As long as I understand your explanation, it is not ideal to write टि्प (dTta+dSignI+dHalant+zero width non joiner+dPa), like this in bellow screenshot

we should may write ट्‌पि (dTta+dHalant+zero width non joiner+dPa+dSignI) like this in bellow screenshot

in this case I assume that I may understand your this point.

For now if we just forget about dUdatta and only focus on above writing style.

I am still confused that what if i use your font and type ट्पि (dTtPa+dSignI) !, in this particular case, is this correct/ideal as modern writing style in Devanagari? Chrome may not support your actual font so I am sharing screenshot bellow so we can understand it clearly,

I typed this in your TiroDevanagariHindi 1.62 font, is this above writing style is ideal as modern Devanagari?

I am asking because I just want to substitute of dTtPa to dTtPa.ss02 in my modified font and this dTtpa.ss02 will look like this as bellow screenshot.

It is a single glyph, it is visually looks like two different character, but cursor in MS Word can not behave like these are two different characters. Is It look correct/ideal as modern Devanagari?

I am asking because my intention for to create dTtPa.ss02 is, in my transcription work I am working on Jain Digambar Shastra, and there are many words which need to visually look like my above screenshot (visually need dHalant between two characters), so I just want to create substitutes for all those glyphs in your 1.62 version font.

I am not quite sure that I will also may need to add dSignI or dUdatta for these substitutes or not for now, but for now I just want to be clear/sure that replacing dTtPa to dTtPa.ss02 as I shown in above screen shot is correct or not?

And if typing this (dTtPa+dSignI) in your 1.62 version font is correct/ideal which like bellow screenshot

so if above sequence is correct/ideal, then what your opinion about if I type dTtpa.ss02+dSignI in my modified font? I am sharing screenshot bellow, because it is just a substitute glyph and behave like one single glyph/character in MS Word, I just modify your font, so it will be easy to understand both of us.

I am stuck here at this point and I don't understand how it is that important if I create a substitute for dTtPa to dTtPa.ss02, it's just visually look different but Unicode is still behave like this is just a one character of two character's combination, there will be no two separate characters as Unicode if I use my above dTtPa.ss02.

I like your font very! very! very! much, because I was just finding font like this which characters are visually similar like native Devanagari/Hindi writing style on paper with pencil. There are many different shaped fonts but those are much complicated for modify, and your font is visually almost look nice and also have not that complex logical need to modify it.

This 1.62 font still have some problems that I noticed, and I want to share those problem here with you, but for now in my modifying process I feel that first I need to solve my this doubt about this substitute of dTtPa to dTtPa.ss02.

Once your concept about this will be clear in my mind, very first I will make all those substitute glyphs, and then my second stage will be on other issue in this font.

I heartily appreciate your every reply even if I don't understand English that much, but still we are working and I hope we will be continue working on this.

I am asking this much long-long and it's may sounds like I am selfish and asking solution just because I want a solution of my doubts and you are the only one source I found that is reliable, but it's not 100% true, I like your font very much and want this font will be much better and better and have options for users like me.

Again thank you so much tiroj. I understand it may make you feel tired to explain concept repeatedly every time which you understand about knowledge of Devanagari in your life to me.

All I can say is THANK YOU SO MUCH for creating this font and replying me. Gratitude... Gratitude...

[tiroj]

This is the traditional form of the TtPi conjunct, which we favour in our fonts because it is possible to use ZWNJ to prevent the ligature, but not vice versa, i.e. there is no way to override an explicit halant using a formatting control character. So we make the traditional form the default. Also, the Tiro Indic fonts developed out of the typefaces we made for the Murty Classical Library of India, so the traditional forms were preferred.

If you use ZWNJ to force an explicit halant, then you are going to get this form

because the shaping engine that looks after the ordering of the glyphs follows modern Hindi conventions and does not reorder an ikar past an explicit halant (in the other direction, it will also not reorder a repha past an explicit halant.

There is a way to build a font such that the default display of TtPi is what you want, but it means changing the default behaviour. It sounds like this might be what you want, given your transcription needs: a custom version of the Tiro font with modified layout tables that displays TtPa — and perhaps some other conjuncts? — using a nominal half form substitution, i.e. Tt with a halant as a combined glyph instead of having an actual halant glyph in the text run.

The Tiro Devanagari Hindi source actually contains the glyphs you would need for this — I included them in anticipation of this kind of request —, but they are not currently used. So there is a glyph in the font called /dTth.half, and similar glyphs for other letters that do not take actual half forms:

So it would be possible to fork the font under some new name, and edit the OpenType Layout tables to change the way the conjuncts involving these letters are processed. This would involve

1. Adding these to the half form substitition lookup
2. Removing the traditional ligatures in the cjct substitution lookup
3. Adding these half forms to the relevant half form groups used in ikar width selection etc.

I think that is probably all that one would need to achieve your desired presentation.

[mehulchanpura]

Thank you so much tiroj.

I may not understand your last part where you mention this **'So it would be possible to fork the font under some new name, and edit the OpenType Layout tables to change the way the conjuncts involving these letters are processed. This would involve

Adding these to the half form substitition lookup
Removing the traditional ligatures in the cjct substitution lookup
Adding these half forms to the relevant half form groups used in ikar width selection etc.'**

But I understand and I feel that it will be not a good idea to create substitute of any glyph which substitute need to visually look Halant between two characters. Because in future if I will need to add dSignI with that newly created .ss02 glyph, it will may not visually look ideal, so I am thinking not to create substitute glyphs for those character combinations which may need Halant between them.

Even if I use zero width non joiner for it, and it will be visually look ideal if I will need to add dSignI after last character.

What I understand for now is this.

So I feel that it's better to use zero width non joiner instead of creating new substitutes for those glyphs.

Now I have new doubt about to create substitute glyph which may doesn't need Halant between combination of those two characters.

There are some glyps for example like dKNa, I created a substitute glph for this named dKNa.ss02, because my pdf visually look different for combination of this both characters, here I am sharing screenshot of both, first one is your original font combination and second one is my created dKNa/ss02 glyph, so you can visually understand what I want for my document.

There I also add dSignI in screenshot, because I already modify dSignI default to wide for this substitute in font lookup.

Is this correct to create substitute of this types of glyps which doesn't need Halant between those?

If you want I will generate a new issue for this or if it is okay for you, we can continue here.

Again thank you so so much tiroj, because you guide me how Halant between two characters need attention.

Thank you for your kind guidance.

[tiroj]

But I understand and I feel that it will be not a good idea to create substitute of any glyph which substitute need to visually look Halant between two characters. Because in future if I will need to add dSignI with that newly created .ss02 glyph, it will may not visually look ideal, so I am thinking not to create substitute glyphs for those character combinations which may need Halant between them.

To clarify, my proposed approach does not require creating .ss02 or other new glyphs. Rather, it changes the behaviour of the retroflex consonants and others that take a visible halant in horizontal conjuncts to act as if they were half forms. The ikar width selection in this case is triggered by the same contextual lookups that apply to half form sequences.

Regarding the second part of your message: for many combinations not involving visible halant, you can trigger a horizontal layout like this by using ZWJ U+200D after the halant character. This triggers the half form of the first consonant, and overrides the traditional ligature form. So, for example, this form

can be rendered with dKa dHalant ZWJ dNa dSigni क्‍नि

[mehulchanpura]

Thank you so much tiroj for to remind me that I can use ZWJ also, but if I will use in this 1.62 version dSignI maatra will be implied same as dSignI, instead of if combination of two or three character will should be implied or replace dSignI to dSignI.W.

This is the first part of my reply/message.

I am using chrome browser and in your reply I can see that dKa+dHalant+ZWJ+dNa+dSignI is still looks perfect, maybe because chrome use different font, I am just sharing screenshot of your reply first, and then I well share other screenshot of text table which I created to explain what is the problem if I use ZWJ instead of creating new substitute.

First I want to share a screenshot of your last line of your reply which contains 'क्‍नि', it's looks perfect in chrome browser, but in document editor like MS Word or any other if I type this dKa+dHalant+ZWJ+dNa+dSignI, it look different, so please first check this screenshot of our conversation here in chrome browser.

In your reply while you type 'क्‍नि' in this chrome browser, it's look okay because may chrome's font use correct width of i maatra for this combination.

But in our case, if we use this combination dKa+dHalant+ZWJ+dNa+dSignI, this will not replace dSignI to dSignI.W.

I prepare one text table so we can understand it better, and I am sharing screenshot of that text table bellow.

As I told I don't have idea how to manipulate logics, like if we use ZWJ. And I already created few substitutes for this types of character combinations (glyps).

So in first column I just type as you suggest me like 'dKa+dHalant+ZWJ+dNa+dSignI' and in second column shows the combination of substitutes which I created in my modified font. In second column I just type 'dKNa.ss02+dSignI', 'dGNa.ss02+dSignI', 'dGhNa.ss02+dSignI'.

I already add these types of few substitute glyphs which I created as per my requirement in 'lookup 11 contextual 2', so when I type above substitutes with dSignI, it will visually display like above screenshot. I mean with my created substitutes with dSignI will be replace dSignI to dSignI.W, and as I feel it look better then we can use 'dKa+dHalant+ZWJ+dNa+dSignI'.

My understanding about font logic like how and which logics need to create GSUB & GPOS is very limited, but I create substitute for those, just because I think this will be better instead of using ZWJ, because i maatra is not changing in this native font and as per my expectation it should be like dSignI to dSignI.W if glyph length is changed to wide. Maybe there is a solution for this also available in our native 1.62 font! if there is please guide me.

So for nor now I have this doubt, and I want to ask you is this correct if I need these types of substitutes in my transcription text which also need dSignI.W ?

This bellow is the second part of my reply.

There is lot of work I need to do with your font if I want to modify as per my requirement, and also there is some core concern I noticed and I am very hopeful that we can solve this. I mean you and your team can solve this. I canl just give you feedback or ask my doubts about this font, rest is totally depend on your and your team.

Me and rest of all other users are depending on you only.

tiroj thank you very very much again. I hope this font will be the best and simplest Devanagari Unicode font in word.

I appreciate you and your team's hard work to build this font.

This is the third part of my reply/message.

For now I am focusing on how can I modify this font as per my requirement, but I want to go step by step, I have other concern also about this font, but later we will can discuss on that. Because if I start sharing all of those my feedbacks about this font to you, it will maybe to complicated for me to understand your concept about this font logic.

So for now I am sharing or asking my doubts which are related to my desired font, but I promise I will feedback you all those things which I noticed, for example dChandrabindu is not positioned visually look nice in some cases, but for now I am avoiding this, because I can not learn or understand everything all at a time.

So please don't think like I want only about my desired modified font. I will be here and after solving this issue, if you want, we will be generate a new issue whenever you feel to do this.

Since that, again thank you so much.

My English is poor so I want to tell you in Hindi, धन्यवाद! मैं कृतज्ञ हूँ कि मुझे आपसे मदद मिल रही है।

🙏🙏

[tiroj]

The context statement in the lookup that substitutes the widest ikar variant glyph can contain sequences of glyphs or groups of glyphs, not just individual glyphs. So I think it should be possible to automatically trigger this substitution when a half form sequence has been triggered by ZWJ. In fact, this is probably what should be happening by default in these Devanagari fonts.

At present, the context statemement in the lookup dxxx.pres_ikar\W is

| <x.ikarW>

we can add a second statement, so the context is

| <x.ikarW>
| <d.half_forms>

That will susbtitute the wider ikar before any half form sequence (the <d.half_forms> group already exists. I think this will be appropriate, but needs to be tested with the narrowest half forms followed by the narrowest full forms to make sure the ikar is not too wide.

It is, of course, possible to add variant conjunct ligatures as you propose, but it may not be necessary to achieve the results you want. A lot of the decisions depend on how you want to work with the font in your document process.

---

Regarding the chandrabindu positioning (and also the anusvara positioning): this is based on our review of manuscript practice and what was done in the best metal types. A lot of technologies, from typwriters to digital fonts, have positioned these mark always above the final stem of the base letter, but that is not the traditional scribal practice.

[mehulchanpura]

Thank you very much tiroj.

I try your this your bellow suggestion

| <x.ikarW>
| <d.half_forms>

and it works, as I believe it is good to use inbuilt options in a font if possible instead of creating substitutes. I completely agree with it, and even if this will be much easier for me to use this font for my transcription.

This is the first part of which I understand form your guidance.

This part in your last reply,

'That will substitute the wider ikar before any half form sequence (the <d.half_forms> group already exists. I think this will be appropriate, but needs to be tested with the narrowest half forms followed by the narrowest full forms to make sure the ikar is not too wide.'

After reading this part of your reply, I was making a list of all those dx.half or dx.halfx glyph's width, I found that narrowest width in this glyph named dG.half and it have 307 width, which is narrowest width of half form in our font as I can understand.

Here I have more then one questions, first one is suppose if I want to type dG.half+ZWJ+dGa+dSignI, for this sequence I may need to add dG.half glyph in to Lookup 11 contextual 1, which is followed or assigned to replace dSignI to dSignI.M. I already do this and it's look good, but my concern is what if there is or will be a next glyph after ZWJ which have wide width? for example dG.half+ZWJ+dKh+dSignI or dG.half+ZWJ+dMa+dSignI, because both dKh and dMa is wider then narrowest dGa.half.

In this type of situation can we able to insert or add any logic in font or we only must need to create a substitute glyph for those character/glyph combination only if we want desired visual look combination of characters which had used ZWJ and also have dSignI (ikar) at the last ?

Second part; (Second question) of my message is I noticed that there is a two different glyphs named are dCh.half and dCh.halfalt. I was trying many ways to type that dCh.halfalt form with different character combination, but still I don't know how can I use this dCh.halfalt if I want to type dCh.halfalt+ZWJ+dVa or any other second or last character combination.

Because in my transcription script/pdf this combination of characters visually look different, and that's why I create a substitute for to show you that what actually I need. I am sharing screenshot of both first one is our original font and second one in screenshot is my modified(created substitute for dChVa to dChVa.ss02) or my desired combination of character which I want to visually look like this and for that, either I need to learn how I can use dCh.halfalt+ZWJ+dVa ? or I will need to create a substitute for this glyph like I already created for to as a trial. Bellow screenshot is just showing how our font visually look like if we type dChVa.

But I want this combination will be look like bellow screenshot.

Bellow screenshot which I created substitute using dCh.halfalt's shape and positioned dVa, which may look very similar with my desired look. Bellow is the screenshot which I created for to show you that what visually I want.

Is that possible that I can use this dCh.halfalt whenever I will need to use with ZWJ or some how other way?

I may not that good to explain my doubts or questions in English that well, and that's why you may feel like repeating or insufficient explanation of my doubts. But thank you very much to teach me that how can I use ZWJ and add glyph in contextual lookup.

I am hoping that you may not get tired of replying.

Thanks a lot.
God bless you & all.

[tiroj]

Here I have more then one questions, first one is suppose if I want to type dG.half+ZWJ+dGa+dSignI, for this sequence I may need to add dG.half glyph in to Lookup 11 contextual 1, which is followed or assigned to replace dSignI to dSignI.M. I already do this and it's look good, but my concern is what if there is or will be a next glyph after ZWJ which have wide width? for example dG.half+ZWJ+dKh+dSignI or dG.half+ZWJ+dMa+dSignI, because both dKh and dMa is wider then narrowest dGa.half.

It might be desirable to add some glyph groups in which full forms letters are grouped by their width, so e.g. <d.full_forms.M> (medium or narrower). Then you can have a context string in the lookup for the M ikar:

| <d.half_forms.N> <d.full_forms.M>
| <x.ikarW>

i.e. a narrow half form followed by a narrow or medium full form.

The lookups for ikar width selection are set up to be processed as subtables, which means that if a match is found for the substitution input in one lookup, then the subsequent lookups are ignored. So you would keep the contexts

| <x.ikarW>
| <d.half_forms>

in the wide ikar substitution, and this would pick up any half form sequences not already used to trigger the medium ikar.

[tiroj]

(Second question) of my message is I noticed that there is a two different glyphs named are dCh.half and dCh.halfalt. I was trying many ways to type that dCh.halfalt form with different character combination, but still I don't know how can I use this dCh.halfalt if I want to type dCh.halfalt+ZWJ+dVa or any other second or last character combination.

/dCh.halfalt is one of the things I included in the font in case it was desired in future, but don’t actually use in the font lookups. Some of these things are there in case someone wants to make a version of the font that works with the older deva shaping model that was used before Windows Vista, or in case someone wants different default behaviour.

A lot of Devanagari text is written without a half form of the Cha, and it is treated like the retroflex letters and written with an explicit halant. So that is the default handling in the font.

If you want to activate the /dCh.halfalt form and have it used in half form sequences with ZWJ, I suggest going into the half feature substitution lookup and adding the substitution

dCha dHalant -> dCh.halfalt

at the appropriate place.

[mehulchanpura]

Thank you tiroj.

I understand your second (last) reply that how to activate dCh.halfalt in lookup, I tested it and it works. But for now I don't want to dive deep in to this dx.halfalt feture.

For now I want to focus on your answer about my first doubt, I read that many times and try to understand actually how to do that. Honestly I don't understand meaning of your explanation.

I don't understand what you mean by "It might be desirable to add some glyph groups in which full forms letters are grouped by their width, so e.g. <d.full_forms.M> (medium or narrower)." I don't understand, do you mean that we need to create more new glyphs for those combination like dx.half(Narrowest)+dxa in font? Because if we need to create whole new glyphs for to use ZWJ it will be many-many glyphs we need to create. I check and I find that I may need 30 to 35 substitute glyphs for our original 1.62 font's glyphs for to get my desired visual look of my text.

If your meaning for "add some glyph groups" is not that I explain that what actually I understand from that part of your reply. If your meaning is as I understand then can you please explain it how can we do that? and if your meaning is different from which I actually understand, in this case can you please reexplain your whole reply in little simple way (at the level of my understanding about modifying font, and I am just a beginner) so I will may understand.

I know it's tiring to reexplain, so I am totally leaving it on you, whatever you choose I will be grateful for everything I learn from you.

And then I also not understand this part of your reply
"| <d.half_forms.N> <d.full_forms.M>
| <x.ikarW>
i.e. a narrow half form followed by a narrow or medium full form."
I don't understand this whole thing.

I don't understand this also
"The lookups for ikar width selection are set up to be processed as subtables, which means that if a match is found for the substitution input in one lookup, then the subsequent lookups are ignored. So you would keep the contexts

| <x.ikarW>
| <d.half_forms>
in the wide ikar substitution, and this would pick up any half form sequences not already used to trigger the medium ikar."

Maybe this is may reason that I have not any type of programming language knowledge.

I don't understand what to do now, still keep trying understand that how can we use ZWJ even if combinations or characters with also have dSignI (ikar matra) or to go ahead with creating those substitute glyphs in font which I will need to create for same visual look of my text similar as my pdf?

I really appreciate your all efforts to explain solution for my doubts.

Thank you so much for your kind approach and efforts.

[tiroj]

By adding glyph groups, I mean defining named lists of glyphs to be used in the substitution lookup context strings. So, for example, <d.half_forms> is an existing glyph group (some tools refer to these as ‘classes’, rather than ‘groups’) that contains a list of all the half form glyph names. What I am suggesting is creating new glyph groups that contain lists of the narrower half forms and narrower full forms, which can then be used in the medium ikar selection lookup context like this:

| <d.half_forms.N> <d.full_forms.M>

That context string tells the font to substitute the medium ikar variant when followed by a sequence of a narrow half form and a narrow full form.

Regarding subtables, this is a particular kind of structure in the OpenType GSUB table that determines how lookups are processed. In a subtable structure, one can have a grouping of multiple lookups that are processed together. What this means is that when a text string includes a mach for a substution in one of the lookups, it performs the substitution and then ignores the remainder of the lookups. This is a useful structure for processing exceptions, so what I am suggesting is that the final lookup for the widest ikar substitution can have a general context rule

| <d.half_forms>

while the lookup for the medium ikar substitution can have an exception to this for the narrower half forms when followed by narrower full forms.

In the VOLT project source, the backslash character \ in a lookup name indicates that it and adjacent lookups with the same name to the left of the \ will be compiled as a subtable structure. So the three lookups

dxxx.pres_ikar\N
dxxx.pres_ikar\M
dxxx.pres_ikar\W

are processed as a group.

[mehulchanpura]

Thank you tiroj for still trying to explain even knowing that I am just a beginner. Heartily thanks a lot. But it's still not that easy for me to understand what you mean by that in your explanation, and if we will go ahead with this, it will may time consuming for you, like everyday you will may need to explain or teach me how to do what you were explained in your reply.

So please don't bother if you feel like discontinue, you may close this conversation. If you feel it will be tiring, much time consuming or wasting of time for you. Whatever decision you will take, I will be always grateful for it.

I am using last latest version of fontforge for font modification in my laptop which have Windows 10.

Here I want to tell you about my knowledge about modification of Devanagari Unicode font (Same as my native language Gujarati, I tried to modify Gujarati Unicode font also in past, but now I am focusing only on your this 1.62 version of Devanagari Unicode font), I am and I was just tried to change/modify font like remove something from any existing lookup to know is it working or not! for just as an experiment, but whole concept or logics behind how every lookup works! I feel that I am still not that able to understand that all things.

That time I was just try to learn with experiment, but I was not fully satisfied, because I have not that logical knowledge or it is may very complex logic for my me.

So I just tried to understand how can I add or remove any glyph from any existing lookup so I can get my desired visual result while I am typing.

For this I tried many different experiments, I created substitute glyphs or remove any existing glyph so I can get desired result, but as I told I was not satisfied with that.

The other thing I tried was to create new Single Substitution lookup, and I was successfully created that new Single Substitution lookup with the help of ChatGPT, Then I tried to create new Contextual lookup or add that Single Substitution lookup in to existing Contextual lookup, but that time, maybe because lack of my knowledge I didn't understand why my modification in Contextual lookup isn't working.

That time I tried ChatGPT for to know about the solution of my doubt, I spend more than two weeks and honestly from ChatGPT I know how can I create new Single Substitution lookup, but at the end I feel that ChatGPT's free version have limited number of process for free users, and ChatGPT says me that it need to process our/my .sdf file to use fontforge in it's own server environment and that processes count is more than my free daily limit, and because of that I wasn't able to get result if I upload our .sdf file to ChatGPT, because after uploading ChatGPT need to use fontforge in it internal environment, and I am a free user so every time ChatGPT reply me your daily limit is over. So I though it's better if I will direct contact font developer, and then I raised my issue here.

But while when we are discussing, I found that there are many things that I really don't need to do. For example I was creating Substitutes of those glyphs which characters combination need to be look visually separate (like dGNa to dG.half+ZWJ+dNa, for example if I want to type अग्नि, our font we already have glyph for combination for this like dGna, and if I want to type अग्नि word using with default glyph will be look like this without using ZWJ, as bellow schreenshot,

but my pdf visually need to type अग्नि look as like this as bellow screenshot, and for to get this visual result I created a substitute of dGNa.ss02 for dGNa and after creating this substitute I added this glyph in Contextual 2 lookup which represents dSignI to dSignI.W will be look good using this substitute, like bellow screenshot

particularly in this case if I use ZWJ then dSignI maatra will be as default dSignI instead of dSignI.M or dSign.M as bellow screenshot,

it is just an example why I get interest to know how can we use and modify font if we want to use ZWJ instead of creating new substitute.) for my desired pdf's visual, but you suggest me an option like ZWJ that how we can do this without creating new substitute glyphs.

This is the first part of my message :- Now I am coming to the point of your explanation how we can achieve our goal without creating new substitute glyphs, even if we have option to use ZWJ.

And for that we are discussing it last days, but I may not understand what actually I need to do. So Now if you are still willing to guide me, I want to go step by step. Please don't bother if you don't have that much free time or interest, you can just close this conversation without any bother.

As you suggest me this:

"What I am suggesting is creating new glyph groups that contain lists of the narrower half forms and narrower full forms, which can then be used in the medium ikar selection lookup context like this:

| <d.half_forms.N> <d.full_forms.M>"

I understand it as, I need to create new lookup in GSUB for narrowest half form, and I created it as 'half.N' Devanagari2 lookup 64, then I add subtable and named that 'half.N' Devanagari2 lookup 64 subtable, and then for now I just add only dG.half in a subtable, same like 'half' Half Forms in Devanagari2 lookup 7 which is already exists in GSUB.

I successfully add this lookup or as in your language glyph group. I am sharing screenshots one by one of my fontforge's GSUB so you may visually understand what I did.

I may able to create this new lookup/group, because there is already a similar lookup already exist in GSUB, so I just observer and study that how can I create this type of new lookup! and because there is already a similar lookup I may copy-paste that logic in my new lookup/group.

I don't know my this newly created lookup or group of glyphs is correct same as you explained or same as you expected or not! If the way I create is not correct, please very first guide me how can I create a new group of glyphs for narrowest half form? And if my created lookup or group of glyphs is correct then we can go to next stage/para of our conversation, otherwise it will be very difficult for me to understand multiple things at a time. so if I was wrong in creating new group of glyphs or lookup in GSUB, please I urge to you that very first guide me how can we create it in my machine environment (using fontforge in Windows 10 machine)? And if this created new group of narrowest glyphs is correct, we may go to next level.

This is the second part of my message :- I don't know and don't understand that also that how can we create a new lookup/group of glyphs for 'Full form'? I don't understand it because I search here & there in GSUB lookup for to full form, but I didn't get. So I am confused about it.

Very first how can I create new lookup/group for full from in GSUB? and then how I can create or define full form of any character in GSUB?

I understand that dG.half is need combination of two different glyphs like dGa+dHalant, but I don't understand how can I define full form of any full form! do may I need to define dGa in full form like dG.half+dSignAa???

I really don't understand whole concept, so I want to go step by step. And if possible please guide me or share a screenshots steps step by step of GSUB, so I can understand logic behind every steps.

Most probably you may use Mackbook/IOS and I am using Windows 10, and also maybe you and your team create this font using different software instead of fontforge, but for now I have a Dell laptop and Dell AuroraR13 Desktop with Windows 10, so it will may not that easy for you to sharing screenshots of those steps using different OS or software, but I have still remain little hope that I will may get solution of this tomorrow or in future.

Thank you so much tiroj, I understand that it will be not that easy for you to come down to my level of knowledge and explain or teach me how we can do this! I heartily appreciate your every single effort whenever you reply me.

Maybe most of the texts in my reply is may have emotional touch, I know it. It may very boring for one professional person, but I can not stop me to write like this, I just want to explain everything that what I feel while passing this phase where I don't have that knowledge and you are still trying to solve and explain the solution of my doubts. And that's why my reply contains this types of texts, you can ignore this.

I have not any mentor and when my knowledge is very very limited, AI also can not help me, and then I ask my doubt to you, and then you reply of my every single message.

And that's why I repeatedly show my gratitude to you.

Thank you so much again.

[miloush]

I just want to say that I am following this discussion and find it very useful, thanks @tiroj for helping out!

[tiroj]

Let me begin by saying that I am not at all familiar with Fontforge, and mostly hear bad things about it even from people who are committed to open source and free software. Unfortunately, there does not yet seem to be a good free general purpose font editor, although there are lots of good free libraries like FontTools for post-design production.

I am presuming that Fontforge uses the Adobe AFDKO feature file syntax code model for OpenType Layout. That’s fine, but I don’t use it except for simple projects. As explained in the Indigo repo readme, the OTL tables in the Tiro Indic fonts are developed using Microsoft’s VOLT tool, and then merged into the final fonts during the build process.

[The VOLT tool is free, but not open source. It only runs on Windows, is almost 30 years old, and was programmed in Visual Basic. I think it is still the best graphical editor for OpenType Layout data.]

If you open the compiled fonts in a font editor, you will be seeing feature file syntax based on decompiling of the font tables, however the tool presents those, and this may not exactly correspond to the input in my VOLT projects, and will definitely use different lookup and group names since those are not available from the compiled font.

[FontTools contains a volt-to-fea converter that you could use to ensure that your OTL code in Fontforge corresponds to the VOLT project.]

---

_"What I am suggesting is creating new glyph groups that contain lists of the narrower half forms and narrower full forms, which can then be used in the medium ikar selection lookup context like this:

| <d.half_forms.N> <d.full_forms.M>"_

I understand it as, I need to create new lookup in GSUB for narrowest half form, and I created it as 'half.N' Devanagari2 lookup 64, then I add subtable and named that 'half.N' Devanagari2 lookup 64 subtable, and then for now I just add only dG.half in a subtable, same like 'half' Half Forms in Devanagari2 lookup 7 which is already exists in GSUB.

This is a bit difficult for me to follow, because e.g. ‘Devanagari2 lookup 64’ doesn’t mean anything to me, because that is a lookup name that, presumably, Fontforge has applied when it decompiles the GSUB table.

When I say ‘glyph group’, I mean a list of glyph names that is stored in the font GDEF table and is available to lookups as e.g. context. So I am not suggesting that you need to create a new lookup, but that you need to define a new list that can be used in additional context rules for the ikar width selection lookups.

Groups is the VOLT tool term for these lists of glyph names. Other tools may refer to them as ‘classes’. They can be defined in the AFDKO feature prefix like this:

@d.half_forms.N = [dG.half, xxxx, yyyy];

(where xxxx and yyyy are any additional glyph names separated by commas). A group can consist of a single glyph.

You can create a group/class for medium width full forms in the same way:

@d.full_forms.M=[zzzz....]

(where zzzz... is a list of whichever full form glyph names you want to use in in the medium ikar selection context when preceded by a narrow half forms).

---

In summary, I think everything you need to achieve what you want already exists in the font, in terms of the glyphs that exist and the lookups that exist. But I am not sure how or if Fontforge handles some of the things I programmed in my VOLT project, such as the use of subtable structures in the ikar selection. This may depend on what Fontforge uses as a compiler for the feature code: as I recall, Adobe’s compiler doesn’t allow use of subtable structures outside of the kern feature, but the compiler that @khaledhosny updated for fontmake does.

I think you only need to edit the existing lookups using some added groups in context rules.

[tiroj]

I was so busy responding to your text, that I didn’t pay attention to your screenshots of the editing environment in Fontforge.

This looks interesting, but something different from both VOLT and AFDKO feature code, so I am even less able to provide you with simple step-by-step instructions. Hopefully what I have described makes some sense in terms of parallels to functionality in Fontforge.

[mehulchanpura]

Thank you so much tiroj.

I think our working environment or software is different, and maybe because of it I may not understand some of your above reply.

After your latest reply I download Microsoft VOLT in my windows 10, and open our base 1.62 font, but then I really don't understand that environment for how, where and what I need to focus. This Microsoft VOLT environment is totally new for me.

I was trying modify font using fontforge since at least from last 2-3 years. Because of my less ability to understand these types of software that how they works are very-very limited, and if I will shift to Microsoft VOLT, it will may take need much-much time for me to understand how can I work with it and achieve my desired modified font!

When I check your font in my document, I feel that shapes of the characters are almost same as I want, and that's why I want to use this font in my document. Then I feel that I need to add some Substitutes if I want to visually same as my pdf.

But then you suggest that I don't need to create new Substitutes glyphs because there is already an option for this is ZWJ.

Then I start to work on your this suggestion, but because lack of my knowledge and may our working environment is/are different, it will may very time consuming for me how can I achieve my desired modified font using with Microsoft VOLT.

Modification of font is really not my ultimate goal. My ultimate goal is very different. I want to preserve some ancient Jain Scriptures same as they printed in that original form.

And this is not also my ultimate goal, my ultimate goal is I want to transcribe/type those Scripturas in Unicode form. I want to use Unicode font for them, because there are many version of those scriptures are already available in pdf at this time, but those pdf used ASCI format.

And I have strong doubt that how long those individual font creators will may serve to the nation!

I know there are or will options like OCR or Image to text, but my concern is if after converting those ASCI to Unicode if I/We have not that appropriate font which can offer different visual look or options for character's combinations if we want that desired visual look.

So I decided that I will try to convert/transcribe all those Scriptures in Unicode form. Because as long as I know Unicode will be exist as far as internet will be exist in our this universe.

After modification, I want to convert those all Scripturas in Unicode friendly, and I want to create those Books is in other format instead of pdf, because right now even if we create a .pdf book using Unicode font, there are many issue if anybody want to copy-paste texts from that .pdf to his/her notepad or any other Unicode friendly editor.

So I decided that first I will modify font as per my requirements and then I will convert my book to as a webpage using hltm+css+javascript or any other if I feel to need of them.

To be honest, Right now I really don't know that how can I will can be create that file even same as I don't have that much knowledge about html or css or javascript. But that is my ultimate goal. I may use ChatGPT or some other AI for how can I create whole book with my visually desired look as a single webpage, that can be all those facilities that one pdf reader have.

I know it will may took much-much effort and time, but my ultimate goal is that.

There is a font named 'Nirmala UI', this font's all glyphs are totally visually look different from our font, but if I use ZWJ in this font I get this bellow result and I feel that is what I want. I am sharing a screenshot of that bellow,

I still didn't try to study this font, but I want that our this 1.62 font have this facilizes/feature like this, and that's why I was asking solutions for my doubts.

Our working environment are different and for now I am not that able or willing to learn whole new software for modify our font.

Most probably I will create substitutes for those some glyphs in our font and will may go ahead towards to achieve my ultimate goal.

Meanwhile I really heartily appreciate your every single effort.

Thank you tiroj.

If you will solve this issue in your current or next version of font, please notify me, I would really like to use this font for my Scriptures.

Again thank you so so much.

[miloush]

Just for reference, there is some documentation for Microsoft VOLT at https://learn.microsoft.com/en-us/typography/tools/volt/tutorial

[mehulchanpura]

Just for reference, there is some documentation for Microsoft VOLT at https://learn.microsoft.com/en-us/typography/tools/volt/tutorial

miloush, please give a some time to decide either I will learn how to modify font using font using Microsoft VOLT which is totally new for me or I will take shortcut to create new substitutes for those combinations of characters in our this 1.62 font.

Because I am/was an average student and it will be may very difficult for me to learn totally new software for to modify our font.

Now I have to go back to home, but tomorrow I will defiantly try to study your link, and then I will may able to reply you properly, until thank you miloush. I wish I also that able like you all to understand this font concept.

Thank you miloush.

[mehulchanpura]

Just for reference, there is some documentation for Microsoft VOLT at https://learn.microsoft.com/en-us/typography/tools/volt/tutorial

I check your this suggested tutorial link, and try to use VOLT, whatever I understand from this is very little, but It is interesting, but for now I don't have that much time for to learn totally different software.

So for now I am going to create substitutes for those glyphs which I need to be visually look separated. First let me finish my this task using fontforge, and then if I will still may face issue, I will may rethink on it to try this software or not.

I am an average Lerner so I need more time to understand any new software that how it works, specially if there are no video tutorial is available or I only dependent on AI or any forum from where I may get answer or lesson about my doubts.

Thank you very much miloush.

I will may report here or rise an another issue if I will found anything in our this 1.62 version font.

Thank you very much again.

[tiroj]

Apologies for this delayed and brief response: I am in the middle of a project with a looming deadline, so may not be able to respond to this thread very reliably until after 14 July.

I can’t give you specific advice about Fontforge, but I presume that somewhere in that program there is a place where you can define groups/classes, and probably when you open my compiled font the existing groups/classes in the GDEF table will be decompiled to that place in Fontforge. So I suggest looking around for these. You should be able to identify them easily because the existing groups used in the context for the decompiled ikar width selection lookups will appear there. This is where you would add new groups/classes which can then be added to appropriate lookups to refine the contextual rules for the substitutions.

There is a font named 'Nirmala UI', this font's all glyphs are totally visually look different from our font, but if I use ZWJ in this font I get this bellow result and I feel that is what I want.

I also made the Nirmala UI font (for Microsoft), and designed the Devanagari portion. This font handles ZWJ exactly the way in which I am suggesting to you to do so, i.e. with contextual rules for the ikar width selection lookups that include half forms.

This is something that I can do in the Tiro Devangari fonts in future, but I am not able to do it now and am trying to avoid making lots of tiny adjustments to those fonts for public release, as I prefer to work on a lot of issues together and prepare a signficant update across all the fonts.

What I think I could do — after 14 July — is make a custom VOLT project and .input file for you, from which you could compile a local version of the fonts.

[mehulchanpura]

Again thank you very much for replying this thread while you have to do lot of things in your busy schedule. I understand it that one professional have to do much many things everyday.

I appreciate those every reply in this thread/issue.

Until 14 July you may need to focus on your own tasks, I understand it, please don't bother.

Here there are a multiple options I get from your reply.

As you can't give me specific advice about how we can achieve this using fontforge, but you also suggested me some how I can achieve this task even using fontforge. But unfortunately I don't understand that part of your reply. I try to watch and study our 1.62 font in fontforge's menu which represent font's information. I checked GSUB and GPOS, but I didn't find "GDEF table", while I am typing this I tried to search what it's means for "GDEF table" in fontforge on internet. Internet give me a very long reply and I just read that once, but as I said that I have no mentor or faculty which can guide me for fontforge also and because of this I am just is in confusion stage.

I may rise my question to any fontforge forum, but I don't know if there is a dedicated forum for that or not!

Now everything I know about how fontforge works is totally dependent on my online research and based on trial and error.

for now I feel that I need to choose for which software I am willing to go!

This is my first confusion/option :- If I will try to research about how can I do this with fontforge! it will also may take much time like days or may be more then one month to understand that how can I do this which you suggested me for fontforge, in this case I will need to take help of internet or AI, and after all of those my efforts I am not that sure that any AI or internet will be able to solve this type of complex problem.

This is my second confusion/option :- Meanwhile this two days I already created those new substitutes for those glyphs which visually I need in my .docx file. Just today morning I created those substitute glyphs and still didn't check it in my .docx file, so I am not that sure that it will be easier for me or not! I can say that for now I feel that this is the only simple solution of my problem. But here in my heart still there is a question how can I do without creating new substitutes.

This is my third confusion/option :- As you reply me this "after 14 July — is make a custom VOLT project and .input file for you, from which you could compile a local version of the fonts." you give me an option that you may work on it and provide me .input file of your VOLT project. But if I will choose this option, I will must need proper guidance, because I already try to understand or learn from this link "https://learn.microsoft.com/en-us/typography/tools/volt/tutorial" from @miloush and I feel that it is totally new software for me to understand how it is work!

Even with fontforge there are not that much video available in youtube. Just because I want to modify it and then I try to study solely, I took some guidance from AI. I just learned few things about fontforge like how can we modify shape using fontforge's inbuilt tools, and I learned those things which I explained in my last few replies.

Honestly I want to learn this thing about Unicode font, but because I haven't that knowledge and I am little bit afraid that how can I understand or learn how Microsoft VOLT is works!

I am saying this because I already download it my machine which have Windows 10 and I try to follow @miloush 's link, but there I was feel unable to select lookup which is/are already existing in our 1.62 font.

So I am thinking is without core knowledge of any software or without step by step guidance I will may not understand that whole processes which we need while using VOLT!

This is the only reason I may hesitating to go with this option.

I still didn't try AI for how to work with VOLT, but I know AI is not that much able to solve my every doubts where I am not that familiar with Tech Knowledge about Unicode/any font.

All I need to step by step guidance, and without it, it will may take very much long time to working for me to with VOLT.

If @tiroj or @miloush will not be able to guide me step by step that how to use VOLT, I will may not able to achieve our/my task, even if you provide me that ".input" file. Because I really don't know how to work with that file in VOLT.

I am really don't want to pushing you all to that level where you will may need to guide me step by step. All I can promise is, I will try to research here & there on internet or using AI to know how can I work with VOLT, and then if I will find that great person like you while searching here/there I will may learn how to deal with VOLT.

My study about VOLT is very-very limited but I didn't find those options to select any existing lookup or to export my work as a .ttf font.

So These are my confusions, and I really don't know with which I will need to go. I expect that you may understand my level of knowledge, and I have no complain for anything like if you reply me late, I understand that.

Thank you @tiroj and @miloush for giving your precious time for to reply me.

I am and will be always grateful for this.

Again, thank you so much all.