After gathering feedback from several distros and user forums,
I would like to make a suggestion for the fonts-noto package that I
believe will save a significant amount of time for the vast majority of
users, and will provide backwards compatibility with the status quo.

The issue is that the fonts-noto-core package bundles together a very
large number (over 190[1]) of exotic fonts. While the aim of covering
all Unicode scripts (65+) means the package covers our bases for
individual users on every locale imaginable, it also presents a
challenge for font selectors in applications, and for *all* users using
them.

Some applications freeze while trying to render so many fonts.[2]

But more importantly, a typical user is likely to only use *one* of these
language fonts, if any (Ubuntu Desktop statistics show that the majority
of users use the English locale.[3]) Every time a user needs to select
a font, the font selector is cluttered by ~190 fonts they do not need.[4]
Even only 5 seconds wasted scrolling through this list of unnecessary
fonts, and only 1 font selection per day, amounts to several tens of
thousands of person-hours wasted each day, when scaled to the entire
user base.[5]

Uninstalling the fonts is commonly requested[6], but far from trivial[7],
or even impossible[8].

My proposal is simple: split the package in at least two components
that can be installed and removed independently. From what I can see
in the package manifest, the following fonts might have general
applicability:

Noto Sans
Noto Sans Linear {A,B}
Noto Sans Display
Noto Sans Math
Noto Sans Symbols
Noto Sans Symbols2
Noto Serif

These would comprise one package. For simplicity, the other package can
contain all other 180+ fonts. One of them might be useful for users
in certain locales, two for a few bilingual/multilingual users who might
need more than one script, and more than two, only for very specialized
cases (language researchers?).

The meta package can still pull in all dependencies, but this will allow
the vast majority of users to easily uninstall the fonts they don't need.

Note that while I am a software engineer, I am not familiar with
Linux development, so please excuse any potential naivete when it
comes to the arcane details of Debian packaging. I do believe that,
conceptually, the solution I've proposed above makes sense, and I trust
that the ingenuity of Debian maintainers will make its implementation
possible.


[1]: https://packages.debian.org/sid/fonts-noto-core
[2]: https://bugs.launchpad.net/pinta/+bug/1916373
[3]: https://ubuntu.com/desktop/statistics
[4]:
https://askubuntu.com/questions/1140030/how-to-disable-unused-asiatic-fonts
[5]:
https://web.archive.org/web/20170717075850/https://insights.ubuntu.com/about/
[6]: https://askubuntu.com/questions/820746/remove-unused-fonts
[7]:
https://askubuntu.com/questions/214950/how-can-i-remove-fonts-that-i-never-use-from-libreoffice-and-linux-in-general
[8]: https://bugs.kde.org/show_bug.cgi?id=433215

Hi Dan,

Quoting Dan Dascalescu (2021-02-22 00:42:28)

I agree that it makes sense to split the Noto fonts into more packages.

I would prefer to relieve the pain also for _all_ users, however.

Probably makes sense to group by writing system¹ since they are commonly
tied to cultural groups.  I imagine that it is more common e.g. for
users in South India to need other South Indian scripts (and maybe North
Indian scripts as well), than it is for them to need Canadian Syllabics.


¹ https://en.wikipedia.org/wiki/Writing_system


 - Jonas

Agree that a separation per writing system would be best.

What would be the effort necessary to achieve that, compared to "Latin vs.
others"?

BTW, Noto Sans Linear A and B also look like some sort of cuneiform font,
and
I missed Noto Mono and Noto Music from my initial selection of fonts with
general
applicability.

Quoting Dan Dascalescu (2021-02-25 04:43:56)

[snip]

Let me try rephrase:

I will *not* split the packaging of Noto fonts to optimize specifically
for Western society (i.e. Latin + Musical notes + emojis + math bundle).

What I will do instead is generally split more fine-grained - for all
all users globally to be able to mix and match.


 - Jonas

I also support the split. Please split the fonts into their own
packages.

There is a duplicate bug about this:
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=756456

Quoting fabian@greffrath.com (2023-01-19 12:30:48)

The very purpose of Noto is to cover many scripts.

If you need latin-cyrillic-greek coverage, pick another o the many fonts
covering that smaller scope.


 - Jonas

While this for sure sounds noble at first sight, it's also just as
questionable.

I think it's safe to say that the Latin+Greek+Cyrillic variants of Noto
Sans and Serif (i.e. those without a suffix) are considered as general
purpose fonts and *all the others* which carry a specific suffix in
their names are considered as special cases.

I think splitting up the general purpose fonts from the specialized rest
is *at least* the one split that can be asked from this package.
Splitting the more specialized fonts up by "writing system" or whatever
may be left as an exercise. Apparently these splits aren't as "easy" or
unambigious as the aforementioned one, because literally nothing has
been undertaken in this regard since at least two years now.

It seems that this is another case where the perfect (split up by
writing system) is the enemy of the good (split up only the LGC fonts).
Seriously, we need a fonts-noto-latin or similar package, thanks!

  - Fabian

Am Donnerstag, dem 19.01.2023 um 13:12 +0100 schrieb Jonas Smedegaard:

Sure, I was expecting a stubborn reply...

However, you contradict yourself here:

 - Fabian

Quoting Fabian Greffrath (2023-01-20 07:52:49)

I stand by those words, and see no contradiction.

Perhaps it helps (at least others following along here) to include the
sentence that I wrote just before the narrow you made above:


Kind regards,

 - Jonas

Am 20.01.2023 07:59, schrieb Jonas Smedegaard:

A split into "LCG" and "others" would be the first welcome step towards
a more universal split by writing system - and actually the only split
that people ever asked for regarding this package.

  - Fabian

Hello,

Bug #983291 in fonts-noto reported by you has been fixed in the
Git repository and is awaiting an upload. You can see the commit
message below and you can check the diff of the fix at:

https://salsa.debian.org/fonts-team/fonts-noto/-/commit/8e91d366bc0a6cd85f4d89e1f05e7ab6dbbcf2fb
------------------------------------------------------------------------
factor out the Noto Sans and Noto Serif fonts into a fonts-noto-generic package (Closes: #983291)
------------------------------------------------------------------------

(this message was generated automatically)
-- 
Greetings

https://bugs.debian.org/983291

I have created a MR to implement this:

https://salsa.debian.org/fonts-team/fonts-noto/-/merge_requests/1

 - Fabian

Dear Fabian, List,

thanks for packaging fonts for Debian.

I was also struggling with the extensive font list issue. I managed to
deinstall the font today to get rid of the annoyance. Of course, I am
now also rid of a set of type faces.

Why are there so many language specific font files in the package? I'm
not the most knowledgeable person in this field, but, is it not the idea
of a unicode font to be capable to include all languages in a single
font file for one font style / typeface? Then the list of Noto fonts
would slink to only the various styles available (20 or so?).

- Are there technical boundaries that prevent this other than file size?
- Would this be something to discuss with upstream how to tackle this?

I would guess not only Debian faces these problems with Noto?

Another thought would be to split the package out super fine grained,
and make the font installation depend on the language selection of the
users? That would be a lot of work though. I guess there is not an easy
solution, but people are struggling with this font as can be seen by the
various bug reports mentioned in this thread. Especially the fact that a
huge list of fonts is generated by the way the files are structured.

The issue will become harder when more software starts to depend on
Noto, I had both cinnamon and texlive dependencies. Even only the noto
core was too much for my taste...

Best regards,


Tjeerd Pinkert

Quoting dr. ir. Tjeerd J. Pinkert (2023-12-22 17:48:09)

You might find upstrem answer to this question relevant.

If you reinstall fonts-noto-core and less, you can do this:

  zless /usr/share/doc/fonts-noto-core/FAQ.md.gz

...and then read the topic near the bottom by the title "Could you
provide a single font file that covers every language (or at least as
many scripts as possible)?"


 - Jonas

Hello everyone,

I would like to renew interest in migrating the Noto fonts to the new
GitHub upstream sources and splitting the packages as necessary.

https://notofonts.github.io/

https://github.com/notofonts

As each font now has its own GitHub repository, they can be built from the
source code and updated to the latest Git snapshot. The best time to start
this migration would be shortly after the release of Debian 13.

Best,
Amr

[Please do not cross-post]
[Please reply *only* to 1041312@bugs.debian.org]

Quoting Amr Ibrahim (2025-07-11 18:03:16)

Yes, it is still my plan to restructure the fonts-noto packaging now
after the release of Debian 13.

Please do not cross-post. Please reply to this email *only* to the
address 1041312@bugs.debian.org to limit the noise on other bugreports.

 - Jonas

Hi again,

Am 05.11.25 05:08, schrieb Rene Engelhard:
Uploaders: but you are :)

Regards,

Rene