Re: bug#55745: [PATCH] Add support for the Hanifi Rohingya script

Re: bug#55745: [PATCH] Add support for the Hanifi Rohingya script

[Eli Zaretskii]

> From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> Date: Wed, 1 Jun 2022 17:54:24 +0530
> 
> Also, please help me with another problem, I cannot get Emacs to display tamil supplement characters.

With which font?  And what do you mean by "I cannot"? what happens if
you try?

> I have tried adding tamil supplement codepoints to the (tamil) script of (script-representative-chars) function
> in lisp/international/fontset.el

Does the font you use support those characters?  Does it support Tamil
in general?

> I also added the tamil supplement codepoints to the tamil composition rules.

Composition rules have no effect on whether a character is displayed
as a glyph or as "tofu".

> I even added a separate tamil-supplement script to (script-representative-chars) but nothing worked.

There's no tamil-supplement script, though.

  (aref char-script-table #x11fc0) => tamil

See lisp/international/charscript.el, a generated file.

> How can I fix this?

We need first to understand the problem.

P.S. I've moved this discussion out of the bug report to emacs-devel.

Re: bug#55745: [PATCH] Add support for the Hanifi Rohingya script

[समीर सिंह Sameer Singh]

[-- Attachment #1: Type: text/plain, Size: 2254 bytes --]

>
> > Also, please help me with another problem, I cannot get Emacs to display
> tamil supplement characters.
>
> With which font?
>

Thank you for asking the question, I had assumed that it was the Noto Sans
Tamil font, which displayed the supplement characters, but it was actually
Noto Sans Tamil Supplement.

And what do you mean by "I cannot"? what happens if
> you try?
>

When typed in Emacs they are displayed as "Tofu" characters i.e.
Rectangular Boxes with codepoint inside.

Does the font you use support those characters?  Does it support Tamil
> in general?
>

Yes the font supports the supplement characters, but does not support Tamil
in general.

> I even added a separate tamil-supplement script to
> (script-representative-chars) but nothing worked.
>
> There's no tamil-supplement script, though.


Unicode defines one, but AFAIK Emacs removes the "supplement" word from
these characters. I think it is defined in admin/unidata/blocks.awk

On Wed, Jun 1, 2022 at 6:24 PM Eli Zaretskii <eliz@gnu.org> wrote:

> > From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> > Date: Wed, 1 Jun 2022 17:54:24 +0530
> >
> > Also, please help me with another problem, I cannot get Emacs to display
> tamil supplement characters.
>
> With which font?  And what do you mean by "I cannot"? what happens if
> you try?
>
> > I have tried adding tamil supplement codepoints to the (tamil) script of
> (script-representative-chars) function
> > in lisp/international/fontset.el
>
> Does the font you use support those characters?  Does it support Tamil
> in general?
>
> > I also added the tamil supplement codepoints to the tamil composition
> rules.
>
> Composition rules have no effect on whether a character is displayed
> as a glyph or as "tofu".
>
> > I even added a separate tamil-supplement script to
> (script-representative-chars) but nothing worked.
>
> There's no tamil-supplement script, though.
>
>   (aref char-script-table #x11fc0) => tamil
>
> See lisp/international/charscript.el, a generated file.
>
> > How can I fix this?
>
> We need first to understand the problem.
>
> P.S. I've moved this discussion out of the bug report to emacs-devel.
>

[-- Attachment #2: Type: text/html, Size: 3638 bytes --]

Re: bug#55745: [PATCH] Add support for the Hanifi Rohingya script

[Eli Zaretskii]

> From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> Date: Wed, 1 Jun 2022 18:55:17 +0530
> Cc: emacs-devel@gnu.org
> 
>  Does the font you use support those characters?  Does it support Tamil
>  in general?
> 
> Yes the font supports the supplement characters, but does not support Tamil in general.

Then I think this is the problem.  Tamil requires special support from
the font, as evidenced by this snippet from fontset.el:

     (tamil ,(font-spec :registry "iso10646-1" :otf '(tml2 nil (akhn)))
            ,(font-spec :registry "iso10646-1" :otf '(taml nil (akhn))))

As you see, we request special OTF features, and thus by default will
not accept a font that doesn't have them.  You can override that by
customizing the fontset via set-fontset-font.

Are you aware of any good Tamil fonts that support the Tamil
Supplemental block?  I don't seem to be able to find any such fonts.

>  > I even added a separate tamil-supplement script to (script-representative-chars) but nothing worked.
> 
>  There's no tamil-supplement script, though.
> 
> Unicode defines one, but AFAIK Emacs removes the "supplement" word from these characters. I think it is
> defined in admin/unidata/blocks.awk 

Yes, that's what produces charscript.el.

Re: bug#55745: [PATCH] Add support for the Hanifi Rohingya script

[समीर सिंह Sameer Singh]

[-- Attachment #1.1: Type: text/plain, Size: 1831 bytes --]

>
> Are you aware of any good Tamil fonts that support the Tamil
> Supplemental block?  I don't seem to be able to find any such fonts.


After merging Noto Sans Tamil and Noto Sans Tamil Supplement font files and
adding some Tamil supplement codepoints to tamil in
(script-representative-chars).
Emacs is now able to  display these characters.

I have attached the font file for anyone interested.

On Wed, Jun 1, 2022 at 7:02 PM Eli Zaretskii <eliz@gnu.org> wrote:

> > From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> > Date: Wed, 1 Jun 2022 18:55:17 +0530
> > Cc: emacs-devel@gnu.org
> >
> >  Does the font you use support those characters?  Does it support Tamil
> >  in general?
> >
> > Yes the font supports the supplement characters, but does not support
> Tamil in general.
>
> Then I think this is the problem.  Tamil requires special support from
> the font, as evidenced by this snippet from fontset.el:
>
>      (tamil ,(font-spec :registry "iso10646-1" :otf '(tml2 nil (akhn)))
>             ,(font-spec :registry "iso10646-1" :otf '(taml nil (akhn))))
>
> As you see, we request special OTF features, and thus by default will
> not accept a font that doesn't have them.  You can override that by
> customizing the fontset via set-fontset-font.
>
> Are you aware of any good Tamil fonts that support the Tamil
> Supplemental block?  I don't seem to be able to find any such fonts.
>
> >  > I even added a separate tamil-supplement script to
> (script-representative-chars) but nothing worked.
> >
> >  There's no tamil-supplement script, though.
> >
> > Unicode defines one, but AFAIK Emacs removes the "supplement" word from
> these characters. I think it is
> > defined in admin/unidata/blocks.awk
>
> Yes, that's what produces charscript.el.
>

[-- Attachment #1.2: Type: text/html, Size: 2586 bytes --]

[-- Attachment #2: NotoSansTamil-Regular.ttf --]
[-- Type: font/ttf, Size: 628152 bytes --]

Re: bug#55745: [PATCH] Add support for the Hanifi Rohingya script

[Werner LEMBERG]

>> Are you aware of any good Tamil fonts that support the Tamil
>> Supplemental block?  I don't seem to be able to find any such
>> fonts.
>
> After merging Noto Sans Tamil and Noto Sans Tamil Supplement font
> files and adding some Tamil supplement codepoints to tamil in
> (script-representative-chars).  Emacs is now able to display these
> characters.

This is a sub-optimal solution IMHO.  There will always be cases where
a script is split into multiple Unicode blocks, probably mapped to
different fonts.  AFAICS, the Tamil supplement characters don't
interact typographically with normal Tamil characters, so a two-font
solution in this case is valid.  In case my assumption is invalid you
should submit a bug to the Noto project :-)


    Werner

Re: bug#55745: [PATCH] Add support for the Hanifi Rohingya script

[समीर सिंह Sameer Singh]

[-- Attachment #1: Type: text/plain, Size: 1534 bytes --]

>
> This is a sub-optimal solution IMHO.  There will always be cases where
> a script is split into multiple Unicode blocks, probably mapped to
> different fonts.  AFAICS, the Tamil supplement characters don't
> interact typographically with normal Tamil characters, so a two-font
> solution in this case is valid.  In case my assumption is invalid you
> should submit a bug to the Noto project :-)
>

I agree it is somewhat sub optimal, but Emacs does not differentiate
between supplement and "base" characters, so setting different fonts for
both of them does not seem possible.
There also seems to be no font which supports both Tamil and its supplement
characters OOTB.

On Wed, Jun 1, 2022 at 8:00 PM Werner LEMBERG <wl@gnu.org> wrote:

>
> >> Are you aware of any good Tamil fonts that support the Tamil
> >> Supplemental block?  I don't seem to be able to find any such
> >> fonts.
> >
> > After merging Noto Sans Tamil and Noto Sans Tamil Supplement font
> > files and adding some Tamil supplement codepoints to tamil in
> > (script-representative-chars).  Emacs is now able to display these
> > characters.
>
> This is a sub-optimal solution IMHO.  There will always be cases where
> a script is split into multiple Unicode blocks, probably mapped to
> different fonts.  AFAICS, the Tamil supplement characters don't
> interact typographically with normal Tamil characters, so a two-font
> solution in this case is valid.  In case my assumption is invalid you
> should submit a bug to the Noto project :-)
>
>
>     Werner
>

[-- Attachment #2: Type: text/html, Size: 2109 bytes --]

Re: Tamil fonts (Was: bug#55745: [PATCH] Add support for the Hanifi Rohingya script)

[Eli Zaretskii]

> From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> Date: Wed, 1 Jun 2022 22:25:52 +0530
> Cc: Eli Zaretskii <eliz@gnu.org>, emacs-devel@gnu.org
> 
>  This is a sub-optimal solution IMHO.  There will always be cases where
>  a script is split into multiple Unicode blocks, probably mapped to
>  different fonts.  AFAICS, the Tamil supplement characters don't
>  interact typographically with normal Tamil characters, so a two-font
>  solution in this case is valid.  In case my assumption is invalid you
>  should submit a bug to the Noto project :-)
> 
> I agree it is somewhat sub optimal, but Emacs does not differentiate between supplement and "base"
> characters, so setting different fonts for both of them does not seem possible.

Of course, it's possible: set-fontset-font can accept ranges of
codepoints and use different font for each range.  It doesn't have to
accept only script names.

> There also seems to be no font which supports both Tamil and its supplement characters OOTB.

That part is indeed sub-optimal, but not a catastrophe if they are
displayed independently, i.e. there are no composition rules that
cross the boundary between the blocks.

Re: Tamil fonts (Was: bug#55745: [PATCH] Add support for the Hanifi Rohingya script)

[समीर सिंह Sameer Singh]

[-- Attachment #1: Type: text/plain, Size: 1640 bytes --]

Then can we do something like
  (set-fontset-font "fontset-default" '(#x11FC0 . #x11FFF)
                    '("Noto Sans Tamil Supplement" . "iso10646-1") nil
'prepend)

in lisp/international/fontset.el?
After doing this both the "base" and supplement characters are being
displayed correctly.

On Wed, Jun 1, 2022 at 11:09 PM Eli Zaretskii <eliz@gnu.org> wrote:

> > From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> > Date: Wed, 1 Jun 2022 22:25:52 +0530
> > Cc: Eli Zaretskii <eliz@gnu.org>, emacs-devel@gnu.org
> >
> >  This is a sub-optimal solution IMHO.  There will always be cases where
> >  a script is split into multiple Unicode blocks, probably mapped to
> >  different fonts.  AFAICS, the Tamil supplement characters don't
> >  interact typographically with normal Tamil characters, so a two-font
> >  solution in this case is valid.  In case my assumption is invalid you
> >  should submit a bug to the Noto project :-)
> >
> > I agree it is somewhat sub optimal, but Emacs does not differentiate
> between supplement and "base"
> > characters, so setting different fonts for both of them does not seem
> possible.
>
> Of course, it's possible: set-fontset-font can accept ranges of
> codepoints and use different font for each range.  It doesn't have to
> accept only script names.
>
> > There also seems to be no font which supports both Tamil and its
> supplement characters OOTB.
>
> That part is indeed sub-optimal, but not a catastrophe if they are
> displayed independently, i.e. there are no composition rules that
> cross the boundary between the blocks.
>

[-- Attachment #2: Type: text/html, Size: 2317 bytes --]

Re: Tamil fonts (Was: bug#55745: [PATCH] Add support for the Hanifi Rohingya script)

[Eli Zaretskii]

> From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> Date: Thu, 2 Jun 2022 08:25:41 +0530
> Cc: Werner LEMBERG <wl@gnu.org>, emacs-devel@gnu.org
> 
> Then can we do something like
>   (set-fontset-font "fontset-default" '(#x11FC0 . #x11FFF)
>                     '("Noto Sans Tamil Supplement" . "iso10646-1") nil 'prepend)
> 
> in lisp/international/fontset.el?
> After doing this both the "base" and supplement characters are being displayed correctly.

First, you omitted the OTF features we request for Tamil -- aren't
they needed for the Supplement block? if not, why not?

Next, we should consider what the main supported platforms do with
this block -- don't they come with other fonts configured for this
block? if they do, we should at least use 'append', not 'prepend'.

And finally, is the license of Noto Sans Tamil Supplement free enough
for us to include it in our sources?

Re: Tamil fonts (Was: bug#55745: [PATCH] Add support for the Hanifi Rohingya script)

[Eli Zaretskii]

> Date: Thu, 02 Jun 2022 08:38:32 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: wl@gnu.org, emacs-devel@gnu.org
> 
> And finally, is the license of Noto Sans Tamil Supplement free enough
> for us to include it in our sources?
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
I meant "to include its reference", of course, not the font itself.

Re: Tamil fonts

[Visuwesh]

[வியாழன் ஜூன் 02, 2022] Eli Zaretskii wrote:

>> Date: Thu, 02 Jun 2022 08:38:32 +0300
>> From: Eli Zaretskii <eliz@gnu.org>
>> Cc: wl@gnu.org, emacs-devel@gnu.org
>> 
>> And finally, is the license of Noto Sans Tamil Supplement free enough
>> for us to include it in our sources?
>          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> I meant "to include its reference", of course, not the font itself.

According to https://fonts.google.com/noto/specimen/Noto+Sans+Tamil+Supplement/about,
it is licensed under Open Font License.

Re: Tamil fonts (Was: bug#55745: [PATCH] Add support for the Hanifi Rohingya script)

[समीर सिंह Sameer Singh]

[-- Attachment #1: Type: text/plain, Size: 1340 bytes --]

>
> First, you omitted the OTF features we request for Tamil -- aren't
> they needed for the Supplement block? if not, why not?
>

Tamil supplement does not need the akhn OTF feature because it does not
form any conjuncts, they are mostly accounting or agricultural symbols.

Next, we should consider what the main supported platforms do with
> this block -- don't they come with other fonts configured for this
> block? if they do, we should at least use 'append', not 'prepend'.
>

There does not seem to be any font in MS windows which supports these
characters.
I am not sure about MacOS.
but still I will use append.

And finally, is the license of Noto Sans Tamil Supplement free enough
> for us to include it in our sources?
>

The Noto Sans Tamil Supplement font file is under the SIL Open Font License
v1.1
The Noto Color Emoji font is also under the same license and is referenced
in the Emacs source code.


On Thu, Jun 2, 2022 at 11:56 AM Eli Zaretskii <eliz@gnu.org> wrote:

> > Date: Thu, 02 Jun 2022 08:38:32 +0300
> > From: Eli Zaretskii <eliz@gnu.org>
> > Cc: wl@gnu.org, emacs-devel@gnu.org
> >
> > And finally, is the license of Noto Sans Tamil Supplement free enough
> > for us to include it in our sources?
>          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> I meant "to include its reference", of course, not the font itself.
>

[-- Attachment #2: Type: text/html, Size: 2463 bytes --]

Re: Tamil fonts (Was: bug#55745: [PATCH] Add support for the Hanifi Rohingya script)

[Eli Zaretskii]

> From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> Date: Thu, 2 Jun 2022 17:44:47 +0530
> Cc: Werner LEMBERG <wl@gnu.org>, emacs-devel@gnu.org, Visuwesh <visuweshm@gmail.com>
> 
>  First, you omitted the OTF features we request for Tamil -- aren't
>  they needed for the Supplement block? if not, why not?
> 
> Tamil supplement does not need the akhn OTF feature because it does not form any conjuncts, they are
> mostly accounting or agricultural symbols.

What about the tml2 feature, which means OTF-specific support for the
Tamil script?

Re: Tamil fonts

[Werner LEMBERG]

>> Tamil supplement does not need the akhn OTF feature because it does
>> not form any conjuncts, they are mostly accounting or agricultural
>> symbols.
> 
> What about the tml2 feature, which means OTF-specific support for
> the Tamil script?

The Tamil supplement font (at least from Noto) has neither a GPOS nor
a GSUB table.  In other words, no OpenType-specific stuff at all.
It's a simple font with simple glyphs.


    Werner

Re: Tamil fonts

[समीर सिंह Sameer Singh]

[-- Attachment #1: Type: text/plain, Size: 697 bytes --]

>
> It's a simple font with simple glyphs.
>

Yes, Tamil supplement characters do not need reordering, vowel signs,
virama, etc.
so the tml2 tag has no use in this case.

On Thu, Jun 2, 2022 at 6:20 PM Werner LEMBERG <wl@gnu.org> wrote:

>
> >> Tamil supplement does not need the akhn OTF feature because it does
> >> not form any conjuncts, they are mostly accounting or agricultural
> >> symbols.
> >
> > What about the tml2 feature, which means OTF-specific support for
> > the Tamil script?
>
> The Tamil supplement font (at least from Noto) has neither a GPOS nor
> a GSUB table.  In other words, no OpenType-specific stuff at all.
> It's a simple font with simple glyphs.
>
>
>     Werner
>

[-- Attachment #2: Type: text/html, Size: 1222 bytes --]

Re: Tamil fonts

[Eli Zaretskii]

> From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> Date: Fri, 3 Jun 2022 17:07:57 +0530
> Cc: Eli Zaretskii <eliz@gnu.org>, emacs-devel@gnu.org, Visuwesh <visuweshm@gmail.com>
> 
>  It's a simple font with simple glyphs.
> 
> Yes, Tamil supplement characters do not need reordering, vowel signs, virama, etc.
> so the tml2 tag has no use in this case.

This should be in the comments to the corresponding setup of the
default fontset.

Re: Tamil fonts

[समीर सिंह Sameer Singh]

[-- Attachment #1: Type: text/plain, Size: 667 bytes --]

So should I send the patch? I also have the next script ready should I send
it along with this?

On Fri, Jun 3, 2022 at 5:19 PM Eli Zaretskii <eliz@gnu.org> wrote:

> > From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> > Date: Fri, 3 Jun 2022 17:07:57 +0530
> > Cc: Eli Zaretskii <eliz@gnu.org>, emacs-devel@gnu.org, Visuwesh <
> visuweshm@gmail.com>
> >
> >  It's a simple font with simple glyphs.
> >
> > Yes, Tamil supplement characters do not need reordering, vowel signs,
> virama, etc.
> > so the tml2 tag has no use in this case.
>
> This should be in the comments to the corresponding setup of the
> default fontset.
>

[-- Attachment #2: Type: text/html, Size: 1223 bytes --]

Re: Tamil fonts

[Eli Zaretskii]

> From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> Date: Fri, 3 Jun 2022 17:21:42 +0530
> Cc: Werner LEMBERG <wl@gnu.org>, emacs-devel@gnu.org, Visuwesh <visuweshm@gmail.com>
> 
> So should I send the patch?

Yes, please.

> I also have the next script ready should I send it along with this?

Separately, please.

Thanks.

Re: Tamil fonts

[समीर सिंह Sameer Singh]

[-- Attachment #1: Type: text/plain, Size: 860 bytes --]

Great! I will send them in two different bug reports.

I think we should also warn users that if they are trying to change the
tamil font in their init.el, they should use codepoints instead of the
script name,
in the (set-fontset-font) function, because if their font does not support
the supplement characters, they will once more be displayed as "tofu".

Where should I write this warning?

On Fri, Jun 3, 2022 at 5:40 PM Eli Zaretskii <eliz@gnu.org> wrote:

> > From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> > Date: Fri, 3 Jun 2022 17:21:42 +0530
> > Cc: Werner LEMBERG <wl@gnu.org>, emacs-devel@gnu.org, Visuwesh <
> visuweshm@gmail.com>
> >
> > So should I send the patch?
>
> Yes, please.
>
> > I also have the next script ready should I send it along with this?
>
> Separately, please.
>
> Thanks.
>

[-- Attachment #2: Type: text/html, Size: 1484 bytes --]

Re: Tamil fonts

[Eli Zaretskii]

> From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> Date: Fri, 3 Jun 2022 17:46:10 +0530
> Cc: Werner LEMBERG <wl@gnu.org>, emacs-devel@gnu.org, Visuwesh <visuweshm@gmail.com>
> 
> Great! I will send them in two different bug reports.

Thanks.

> I think we should also warn users that if they are trying to change the tamil font in their init.el, they should use
> codepoints instead of the script name,
> in the (set-fontset-font) function, because if their font does not support the supplement characters, they will
> once more be displayed as "tofu".

How is this different from any other font that supports only part of
the script's characters?  Many Unicode blocks have "supplement" parts,
and there's nothing to assure a user that a font which supports the
representative characters will necessarily support all of the others.
Users should be prepared to deal with such problems when they happen,
and there's no way Emacs could warn them about every possible case,
since there are too many fonts out there.

(Personally, I find it strange that a Tamil font doesn't support the
Supplement block, and even more strange that Google created a separate
font instead of extending an existing one.  Perhaps this is because
the Tamil Supplement block is relatively new.)

Re: Tamil fonts

[Visuwesh]

[வெள்ளி ஜூன் 03, 2022] Eli Zaretskii wrote:

>> From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
>> Date: Fri, 3 Jun 2022 17:46:10 +0530
>> Cc: Werner LEMBERG <wl@gnu.org>, emacs-devel@gnu.org, Visuwesh <visuweshm@gmail.com>
>> 
>> Great! I will send them in two different bug reports.
>
> Thanks.
>
>> I think we should also warn users that if they are trying to change the tamil font in their init.el, they should use
>> codepoints instead of the script name,
>> in the (set-fontset-font) function, because if their font does not support the supplement characters, they will
>> once more be displayed as "tofu".
>
> [...]
>
> (Personally, I find it strange that a Tamil font doesn't support the
> Supplement block, and even more strange that Google created a separate
> font instead of extending an existing one.  Perhaps this is because
> the Tamil Supplement block is relatively new.)

Probably because most of the characters added are historic, for a lack
of a better word.  I have never seen them used in practise and I doubt
most Tamil speakers know about them either: maybe if you were old, you
might have used it and/or heard about it [1].  Considering this, I don't
think most font maintainers/creators would have the supplement
characters in high priority list.

Also, these characters were added in Unicode 12.0.

[1] FWIW, even my mother did not know about it.

Re: Tamil fonts

[समीर सिंह Sameer Singh]

[-- Attachment #1: Type: text/plain, Size: 2579 bytes --]

>
> How is this different from any other font that supports only part of
> the script's characters?  Many Unicode blocks have "supplement" parts,
> and there's nothing to assure a user that a font which supports the
> representative characters will necessarily support all of the others.
> Users should be prepared to deal with such problems when they happen,
> and there's no way Emacs could warn them about every possible case,
> since there are too many fonts out there.
>

Ok, so I will not mention that.

(Personally, I find it strange that a Tamil font doesn't support the
> Supplement block, and even more strange that Google created a separate
> font instead of extending an existing one.  Perhaps this is because
> the Tamil Supplement block is relatively new.)
>

I think the noto guys are planning to provide merged fonts to users either
as all the scripts in one font, or on a per-region basis.
See: https://github.com/googlefonts/noto-fonts/issues/2257

On Fri, Jun 3, 2022 at 6:44 PM Visuwesh <visuweshm@gmail.com> wrote:

> [வெள்ளி ஜூன் 03, 2022] Eli Zaretskii wrote:
>
> >> From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> >> Date: Fri, 3 Jun 2022 17:46:10 +0530
> >> Cc: Werner LEMBERG <wl@gnu.org>, emacs-devel@gnu.org, Visuwesh <
> visuweshm@gmail.com>
> >>
> >> Great! I will send them in two different bug reports.
> >
> > Thanks.
> >
> >> I think we should also warn users that if they are trying to change the
> tamil font in their init.el, they should use
> >> codepoints instead of the script name,
> >> in the (set-fontset-font) function, because if their font does not
> support the supplement characters, they will
> >> once more be displayed as "tofu".
> >
> > [...]
> >
> > (Personally, I find it strange that a Tamil font doesn't support the
> > Supplement block, and even more strange that Google created a separate
> > font instead of extending an existing one.  Perhaps this is because
> > the Tamil Supplement block is relatively new.)
>
> Probably because most of the characters added are historic, for a lack
> of a better word.  I have never seen them used in practise and I doubt
> most Tamil speakers know about them either: maybe if you were old, you
> might have used it and/or heard about it [1].  Considering this, I don't
> think most font maintainers/creators would have the supplement
> characters in high priority list.
>
> Also, these characters were added in Unicode 12.0.
>
> [1] FWIW, even my mother did not know about it.
>

[-- Attachment #2: Type: text/html, Size: 3706 bytes --]

Re: Tamil fonts

[Eli Zaretskii]

> From: Visuwesh <visuweshm@gmail.com>
> Cc: समीर सिंह Sameer Singh
>  <lumarzeli30@gmail.com>,  wl@gnu.org,
>   emacs-devel@gnu.org
> Date: Fri, 03 Jun 2022 18:44:44 +0530
> 
> > (Personally, I find it strange that a Tamil font doesn't support the
> > Supplement block, and even more strange that Google created a separate
> > font instead of extending an existing one.  Perhaps this is because
> > the Tamil Supplement block is relatively new.)
> 
> Probably because most of the characters added are historic, for a lack
> of a better word.  I have never seen them used in practise and I doubt
> most Tamil speakers know about them either: maybe if you were old, you
> might have used it and/or heard about it [1].  Considering this, I don't
> think most font maintainers/creators would have the supplement
> characters in high priority list.

Thanks, I think this makes treating them as a separate setting in the
fontset even more reasonable.

> Also, these characters were added in Unicode 12.0.

Yes, thus "relatively new" above.