Possible to add new precomposed characters for local language in Togo?

[Marcel Schneider]

Sorry I see that my e-mail has been altered on the road (“!CRLF ” not found in outbox), I try to send it again (with corrections btw; I was careful, but this time I couldnʼt leave the draft overnight :)

On Tue, 23 Feb 2016 01:53:54 +0100, Mats Blakstad  wrote:

> Exactly where should these combinations be documented? 

In the _Unicode Named Character Sequences_ part of the UCD. Its 8.0.0 version is found at:

http://unicode.org/Public/8.0.0/ucd/NamedSequences.txt


> How much are the difference between if
> people add the diacritics before or after the letter?

Technically, the main difference between entering diacritics before the base letter vs after the base letter appears only if the representation of the target character in NFC is more than one single code unit. And then that difference is only relevant on Windows, not on Linux; as about Mac OS X Iʼve no knowledge. But your question being about XKB (I missed that point), you should get it work by adding the combining diacritics after the base letter on output side in a locally customized XCompose configuration file.

> If people are used to
> add diacritics before the letter, would it not be pedagogically a better
> idea to continue that logic on a new keyboard? What we tried to do is to
> make a keyboard that simply extends the French keyboard (which is by far
> the most used in Togo), and then people can get more keys to a keyboard
> they already know. There are also other keyboards used locally by
> linguistics, but people tend not to learn them, and it can be a barrier
> when people need to click to change keyboard from "French" to a "Local
> languages keyboard" all the time; I guess people prefer a keyboard that
> they can use to write both.

Indeed the use of several keyboard layouts for one single script―Latin script―in the same country is inefficient. To make complete language support widespread, your method of extending the main layout is likely to be the only useful one. A similar effort is actually on-going in France on governmental demand. Iʼm asking for extension to cover all Latin script written languages and translitteration systems, some of what is still an option, beyond full coverage of European Latin script written languages. I feel that people coming from―or studying languages of―countries and communities on other continents should become able to type their language in that script on any computer in France as well as in any other Latin script using countries, given that the issue is only to add some more characters of the same script. Latin script―as opposed to all other scripts AFAIK―stays still chopped into “sub-scripts” and subsets. That brings but counter-productive complications, while the implementation of the whole script is technically feasible even on keyboard level. The only difference between keyboard layouts of Latin script using countries should be varying accessibility depending on frequencies of use.

As about not altering user experience in Togo, on non-Linux/Unix systems I see two options:
1) the keyboard layout on the whole is provided as an IME, ideally in synergy with the French default layout;
2) the dead key handling is carried over to an IME in conformance with ISO/IEC 9995-11, in synergy with a keyboard layout where all dead key characters are combining diacritics―at risk of slightly altering user experience when writing French and the IME is turned off or unavailable.
The first solution has the advantage of being already working, in use, widespread, and fully sustained by SIL, while I canʼt see anything of this in the second solution.

I missed the point that your request is about an XKB layout file. As this obviously targets Unix-based OSes, i.e. Linux, e.g. Ubuntu, output of combining sequences as a result of dead key lead key press sequences is no problem. However combining sequences are actually not found in XCompose as far as I could see on Xlib Compose / X.org web site, but that again may be considered as purely fortuitous.

So on Linux―as opposed to Windows―the goal is a bit easier to reach in that, combining sequences can be generated by dead keys. But it is a bit harder in that, for a given keyboard layout there is no DLL with a complete and customized dead trans list in it. I believe that an installation script could be extended to replace the default composer configuration file by a complete one.

In any case, XCompose needs some thorough update because support for numerous characters used in African Latin script based writing systems is very weak, if not purely missing, in XCompose. Same issue should lead to revise ISO/IEC 9995 again, this year best.

I do hope that your request as well as all equivalent demands from other countries shall be centralized to achieve a thorough overhaul of ISO/IEC 9995 and its international and national implementations to solve at once all pending keyboard problems.

Thanks in turn for your valuable feedback and suggestions!

Marcel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://unicode.org/pipermail/unicode/attachments/20160223/a2ae0508/attachment.html>