The limitations of keyboard layouts (again)

by Michael S. Kaplan, published on 2007/05/04 05:23 -04:00, original URI:

Balkrishna asked the following via the contact link:

hi michael,

saw your blog for the first time today and learnt a lot.

i have just installed MSKLC 1.4 today and tried generating a keyboard layout for a tamil font - non-unicode, typewriter layout.

i read some intelligent comments about indic / tamil kb layouts (yes, inscript was developed by standard setters & software developers as common for all indian languages; convenient for them but not for the user).

in tamil the combination of a consonant and half-vowel has a separate glyph (except for the grantha characters). e.g. 'k'+'u'='ku' = one glyph. This holds good  for all the 18 consonants. I have defined the half-u as a dead character and can generate combined glyph.

the only problem is that the dead character has to be entered first (old old typewriter sequence), whereas since the computers had word processing software, we have been entering 'k' and then 'u'.

Now I would have to enter 'u' and then 'k', which is very confusing when you switch between various keyboarding software.

i could define k,ng,..... as the dead characters and 'u' and 'uu' as the 2nd character. But the 'k' would always be dead, even in combination with aa,o,oo, etc. or one would have to enter space and k (and the other 17 consonants), which is also very confusing.

hope have made myself clear.

any suggestions (if you have time to consider this)? i dont mind a solution where i can edit the klc file and enter hexa codes according to some logic (in my opinion interactive screens and use of mice are bringing us down to the lowest common dumbbell).

basically we would need an option to generate a single code from two consecutive keys without the dead key concept, actually dead keys are not required at all. the keyboard driver has to remember the previous key entered and when the second key of a pair is entered, the driver sends a destructive backspace and the combined code. This logic can be implemented for several-keys to one code too (in hindi/sanskrit one can have 3 consonants with 2 halants in between and a half-vowel/matra at the end, 6 characters in all generating a single glyph).

with best regards

This is a question that comes up often, but unfortunstely it is not possible with the user32.dll-based keyboard model under Windows.

You can have a single keystroke support from one to four UTF-16 code points.

And you can use dead keys (chained or not) to allow two or more keystroke resolve to a single UTF-16 code point.

But there is no other way to do the latter -- more specifically, there is no way that doesn't involve external resources, outside of regular keyboard layouts.

But I will be coming back to the text based TSF TIPs very soon; keep your eye out and you may find the answer to your questions....


This post brought to you by(U+0b8a, a.k.a.a TAMIL LETTER UU)

# Ambarish Sridharanarayanan on 4 May 2007 3:37 PM:

I don't know what it is with Indic languages (and Tamil in particular) and non-Unicode fonts. If a limitation makes life difficult for non-Unicode fonts, more power to the limitation! Switch to Unicode already, people.

P.S. As a native Tamil speaker, I get to install a new font for each Tamil web-site I visit. Soon becomes tiresome, you know?

# Marc Durdin on 4 May 2007 5:52 PM:

Hi Balkrishna,

I think you'll find our product Keyman Developer lets you do what you want - add logic to your keyboard layouts.  Please feel free to contact us for more information -

# Ramesh Muniandy on 26 May 2007 11:57 PM:

I am just a layman. Each time I visit a tamil website, I am required to install certain font. It's irritating.

Tamil people should unite to produce 1 universal font type and 1 universal keyboard layout so that we can open a website with ease.

# Michael S. Kaplan on 27 May 2007 12:43 AM:

If they all use Unicode, then you can use any keyboard you like and you'll already have the fonts. :-)

# Kyoukai on 14 Apr 2008 12:38 AM:

Is it just me, or have I been noticing alot of "tavultesoft" spam?

# Michael S. Kaplan on 14 Apr 2008 10:42 PM:

Spam? Not really, at least not how I look at it.

Marc and I know each other and we know what our respsctive products can do. I don't mind when he points customers at solutions when MS doesn't provide them....

referenced by

2011/04/16 Chain Chain Chain, Chain of Dead Keys

go to newer or older post, or back to index or month or day