asian characters appearing in abrowser without locale package installed

11 replies [Last post]
eric23
Offline
Joined: 06/30/2017

I was wondering if anyone else sees the Asian characters in menus or over images on sites, without having related locales installed. I had the Spanish locale installed, which was a surprise, but I don't that was related.

Here are a few sites, where I am seeing these characters:

http://hoverstream.com/gallery

and

http://civilbeat.org/

Any ideas?

Magic Banana

I am a member!

I am a translator!

Offline
Joined: 07/24/2010

I see no Asian character on Debian 11 with Abrowser 95 and its French localization. Nevertheless, that may be because I have not understood where to look at: could you attach screenshots?

Geshmy
Offline
Joined: 04/23/2015

Using Trisquel 10, Abrowser shows no asian characters at either http://hoverstream.com/gallery or http://civilbeat.org/ for me.

eric23
Offline
Joined: 06/30/2017

Thank you folks. In the first picture you will see faint characters over the pictures of the hovercrafts as well as in the upper right corner. Second picture has some in the menu on top.

I used another computer with Trisquel 10 and it too did not have the same issue. However after posting this it does not come up on computer I post about with screenshots.

My computers always get a little odd. I wonder what these characters actually say.

Screenshot_20220105_175222.png Screenshot_20220105_175506.png
Legimet
Offline
Joined: 12/10/2013

I thought this might be due to Noscript blocking some fonts, but I have NoScript and I do not see those characters. Still, maybe you should try disabling it/other addons and see if that makes a difference.

eric23
Offline
Joined: 06/30/2017

Apparently it is noscript. If you enable font by default in noscript it will show the actual characters. Thanks to everyone.

Connochaetes

I am a member!

Offline
Joined: 12/13/2017

I always see such things. I assume this is what happens when website developers use icons from the Private Use Area (PUA) of a WOFF font they embed in in the website, and then if your browser rejects that font for whatever reason, your browser will instead display whatever glyphs happen to occupy those PUA codepoints in your locally available fonts. Because it's the PUA, those glyphs can be anything. So whether they come out "Asian" depends on which fonts you have installed and how your fonts use the PUA. There are certainly fonts that populate vast expanses of PUA territory with precomposed Korean syllable blocks containing at least one obsolete letter (you can use Unicode's "Hangul Jamo" for obsolete Korean, but I guess providing precomposition via the PUA is a way to achieve the desired look without relying on Jamo composition, just like many fonts come with precomposed diacritics Latin in their PUA, see the ṃ̓/ṃ̓ in your screenshot); or – as in your screenshots – extremely uncommon and therefore not unicodified Chinese characters or variants thereof.
"I wonder what these characters actually say." – In the Korean case, they are random (obsolete) syllables, many of which probably never actually existed before a computer automatically assembled them. In the Chinese case, I guess most are rare glyph variants, names-your-grandpa-made-up-one-day, and "misspellings", all from random old documents.

Avron

I am a translator!

Offline
Joined: 08/18/2020

With noscript, I sometimes see a number of small rectangles with 4 hexadecimal digits, arranged in a 2x2 matrix. I never saw any fancy character.

eric23
Offline
Joined: 06/30/2017

I assume you are using Trisquel 10?

Avron

I am a translator!

Offline
Joined: 08/18/2020

I see that with abrowser on Trisquel 9 and on Debian 11. I haven't really used Trisquel 10 yet although I installed it on one computer.

eric23
Offline
Joined: 06/30/2017

I am assuming you are using Trisquel 9 in the situation where you see the odd characters?

Connochaetes

I am a member!

Offline
Joined: 12/13/2017

It has very little to do with your version of Trisquel (or even MS Windows, or whatever) nor with which browser you use. It would (or wouldn't) occur on every OS, and with many different browsers.

This is what's happening:
1) A website contains special characters (usually, but not necessarily, from Unicode's "Private Use Area" = PUA).
2) Same website instructs your browser to use an external font to display those characters.
3) But your browser is configured to reject said external font, whether directly (for example, enter about:config in the address bar of Abrowser or any other Mozilla browser) or indirectly (by a browser extension configured to block external fonts).
4) So, not accepting the external font, your browser instead uses a local font to display said characters.
5) Especially for codepoints on the PUA, it's likely that your local font will have a completely different glyph than the external font would have had. For example, the external font that the website wanted your browser to use might have had icons signifying "login", "shopping cart", "language", "currency" etc., but your local font has completely different icons on those codepoints. Or it has rectangles showing Unicode's hexadecimal code point. Or it has nonstandard Chinese characters. Or precomposed Korean syllables that aren't covered by Unicode's "Hangul Syllables". What you see depends on what fonts you have or haven't installed. It does not depend on your OS version – only insofar as desktop operating systems usually come with certain fonts preinstalled, and those might indeed differ between Trisquel 9 (Etiona) and Trisquel 10 (Nabia), or between regular Trisquel, Trisquel Mini, and Triskel. For example, I haven't found Avron's "rectangle with Unicode code point" font on a November version of (regular) Trisquel 10, but such fallback fonts exist (Everfont I think? Perhaps Unifont? Noto Fallback didn't seem to have those).

Speaking of fonts, I've recently tried Debian's KDE flavo(u)r, and it comes with a thing to easily define groups of fonts and quickly disable / enable them. This can be useful to easily prune the ever longer list of preinstalled fonts that's a pain to navigate due to the lack of support for navigation by typing the font name's first letter in programs such as regular Trisquel's gucharmap which doesn't even have a scrollbar in its dropdown list of fonts.
I've previously noticed such font grouping utilities on Synaptic, but perhaps a "native" one will work best? So I'm looking forward to try out Triskel in the hopes that it comes with the same utility as Debian KDE.