Using non-latin language on Librem 5

And hereā€™s a screenshot of the Japanese Kana keyboard layout in action with Gnome dictionary (Librem 5 QEMU image):

3 Likes

@dcz I took the time to find the layouts and noticed someone has done a swedish layout (se - shouldnā€™t it be sv as per standards?). It seems identical to the one finns would use for a touchscreen keyboard, so until a better layout manifests, could you possibly copy and use that for the fi layout as well approve the merge request made by @uzanto?
And big thanks to who ever made the swedish layout!

(edit: and concerning https://source.puri.sm/Librem5/squeekboard/issues/117, I suggest using this documentation for a guide for naming layouts/files: ā€œlanguages-sublanguages-regions-ā€¦-special layoutsā€ . Two letter language codes are based on ISO standard 3166 and given additional descriptions if needed (there are lists of languages of the world). Then it becomes identifiable and hierarchical, and additional new layouts (even non-standard) can be added later in a sane way without messing with the main languages/layouts by just adding an additional identifier/code after.)

What about UTF-8 ? It would solve many problems.

UTF-8 doesnā€™t apply to layouts, unfortunately.

If you post here the differences between layouts Iā€™ll make the merge request.

1 Like

No difference with se (which should be sv) touch keyboard. Just copy-paste-rename :slight_smile:
Thanks a lot @uzanto !

(Iā€™ll update the list later).

Waiting for approval

1 Like

Weā€™re using GNOMEā€™s way of naming input methods (the settings application lets people select layouts), so weā€™ll be following that unless it proves insufficient.

But I appreciate the links, they may come in handy in some unexpected moment.

Getting back to topic: do we have anyone working on for instance Chinese, Hebrew, Arabic or Russian inputs? The last one is probably easiest layout wise but still needs a native to work on it.

I created a Russian layout a few weeks back and submitted a merge request but it is outdated now due to breaking changes in the keyboard layout format. Iā€™ve been meaning to resubmit this but Iā€™ve had trouble getting Gnome Boxes to run the qemu image since I updated to Ubuntu 18.04. I hope to be able to get back to this soon.

Good to hear! Iā€™ll mark that as ā€œin the worksā€.

When you select the Swedish keyboard in the Region & Language panel in Settings, the se input source is enabled, so the keyboard needs to use that instead of sv. I donā€™t know why it uses se instead of sv - itā€™s just easier to go with it.

1 Like

If someone gets input for Chinese to work like rinokeros asked that would be great because that would have direct parallels to getting a Japanese input to work that way as well (I believe currently itā€™s just a kind of base kana flick style method?) which is how it works in Linux (Anthy) and Windows.

Right now, the Japanese layout is Kana only with different layers for Hiragana / Katakana and additional layers for numbers and Latin letters. Itā€™s not even flick style, as this is not supported (yet).

For the different missing components/discussions around this topic, take a look at the following issues: #7, #84, #99, #122. The last one is the issue created by rinokeros, but I donā€™t think there has been made a lot of progress yet.

1 Like

I think itā€™s because there is a mixup of country codes and language codes (see https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes compared to https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2). With Swedish and Sweden itā€™s a small difference (to me - not sure if swedes care) but may cause trouble with some other combos, as country code sv= El Salvador, Estonia is ee or et as are Ewe and Ethiopia etc. Problem being that countries and languages are not the same - a common headache with localization (also why two part codes1, 2 may be needed for variants).

It is not uncommon that sv-se is used for Swedish in Sweden and sv-fi for Swedish in Finland (Finland being a country with Swedish as one official language). This combination is necessary for example if a company has different goods in different countries but use Swedish in both.

For a keyboard layout only the language code is necessary and it would be sv according to 639-1.

And there are the Saami in Sweden as well. And Sweden being only one example - Spanish, French, Portugese and several languages without a country.

Just change everything to Esperento or Latin on the phone :slight_smile:

1 Like

Why? Weā€™re all speaking here english, not esperanto, not latin. Better to use de-facto universal language rather than what calls itself ā€˜universal languageā€™.

1 Like

I prefer galactic standard to Esperanto :wink:

I like the idea of 2char language - 2char country across the board personally.

Iā€™ve always found it odd that en is USA English then en-uk is British English. Also having the country code at the end may make sense for financial symbols in some scenarios?

Thatā€™s the problem: we are the anomaly, exception, minority. English is not the most spoken on the globe. And even if itā€™s considered one of the most usable, we have to use other languages for interactions with others. Only a fraction of the globe can use English and for many it gets political fast to be forced to use non-mother tongue - even if they could learn it. An there are hundreds more languages used than there are countries (there are actually thousands of languages and variations and regional dialects) as well as several countries that use same/similar language. Esperanto and latin - used as a joke here - would at least be neutral (mostly) and force everyone to learn something new.

But this was more about there maybe being a mistake in selecting list of codes (country/language) which puzzled me and @david.boddie. I say maybe, because it may have been intentionally selected for some unknown reason or implemented wrong only in this instance or implemented only temporary for this (but if not noticed on some levelnow, it may cause problems later for localization efforts and user experience for the reasons mentioned earlier).

1 Like