Custom font filters for different languages

I'm trying to support multiple languages in a watch face.

I'm using a custom font, and running into memory issues.

In my /resources/resource/resources.xml file I have a filter:

I tried to create language specific filters. E.g. resources-rus/resource/resources.xml includes a filter for russian:

It seems this kind of works, but when the watch face is compiled, only the glyphs defined for the first language (specified in resources-ces/resource/resources.xml) are included from my base font. Then the Czech glyph set is further reduced by the Danish filter (specified in resources-dan/resource/resources.xml) and so on. There are no errors given by the compiler, but in the simulator (and in the actual device) the characters are missing. (For example, in the English translation, "q" is replaced with the rectangular tofu sign since it's not included in the Danish filter.)

Is there a correct way of specifying language-specific font filters for custom fonts?

Top Replies

All Replies

0 jim_m_58 over 6 years ago

If you're using Jungles, I'd look at how you have that set up for resources, and if not, how you have the resource overrides set up for your project.

It sounds like it's not finding the correct resources.

0 peterdedecker over 6 years ago

It's not possible to filter on a per language base. Resources are defined at design time (ie device = fr945) and not at run time (you can switch the language on the device)

0 Travis.ConnectIQ over 6 years ago

I *think* that only string resources are localized using language-specific resource folders, so this is probably by design, but the behavior you're seeing is pretty weird and probably not intentional.

That said, this seems that we should consider allowing font resources to be localized.

0 HermoT over 6 years ago in reply to Travis.ConnectIQ

YES PLEASE!!

0 Travis.ConnectIQ over 6 years ago in reply to HermoT

I'm pretty sure @peterdedecker is right. We won't be able to compile the app with the minimal set of font glyphs because the language is changed at runtime and the fonts need to be compiled into the app.

That said, we should be able to detect a font element in a language-specific resource file and generate a warning if it has a filter attribute. We might also be able to assemble the superset of filtered characters and use that. This would allow you to eliminate characters not used by any supported language, which seems much better than the behavior we provide now.