Dead and Compose keys (was: Re: Romanized Singhala got great reception in Sri Lanka)

Christopher Fynn chris.fynn at
Tue Mar 18 09:46:47 CDT 2014

Hi Andrew

It may be possible with Keyman. I once even wrote a set of MS Word
macros that did the same thing (let users type in Romanized Tibetan
and output Tibetan characters) - however it stopped working when
Microsoft switched from Word Basic to VBA.  :-(

At least Keyman hides all the messy (and poorly documented) details of
Windows system hooks which is what you have to use if you want to make
a stand-alone utility (did that once too).

If Keyman can call external libraries ~ that's interesting. It is
certainly *far* more sophisticated and flexible than MSKLC and I
shouldn't have lumped the two together.

- Chris

On 18/03/2014, Andrew Cunningham < at> wrote:
> Chris,
> Keyman is capable of doing that and a lot more,  but few keyboard layout
> developers use it to its full potential.
> As an example,  I was asked by Harari teachers here in Melbourne to develop
> a set of three keyboard layouts for them and their students.
> The three keyboards were for three different orthographies in the following
> scripts:
> 1) Latin
> 2) Ethiopic
> 3) Arabic
> They wanted all three layouts to work identically,  using the keystrokes
> used on the Latin keyboard.
> The Ethiopic and Arabic keyboard layouts required extensive remapping of
> key sequences to output.
> If I was a programmer I could have done something more elegant by building
> an external library Keyman could call but as it is we could do a lot inside
> the Keyman keyboard layout itself.
> For Myanmar script keyboard layouts we allow visual input for the e-vowel
> sign and medial Ra,  with the layout handling reordering.
> One of the Latin layouts I use,  supports combining diacritics and reorders
> sequences of diacritics to their canonical order regardless of order of
> input. Assuming a maximum of one diacritic below and two diacrtics above
> base character.
> Analysis and creativity can produce some very effective Keyman layouts.
> Andrew
>  On 18/03/2014 7:23 PM, "Christopher Fynn" <chris.fynn at> wrote:
>> MSKLC and KeyMan are fairly crude ways of creating input methods

>> For what you want to - you probably need a memory resident program
>> that traps the Latin input from the keyboard, processes the
>> (transliterated) input strings converting them into unicode Sinhala
>> strings, and then injects these back into the input queue  in place of
>> the Latin characters.

>> There are a couple of utilities that do this for typing
>> transliterated/romanised Tibetan in Windows and getting  Tibetan
>> Unicode output.

>> But I think both of these were written in C as they have to do a lot
>> of processing which is far beyond what can be accomplished with MSKLC
>> and even KeyMan

>> - C

More information about the Unicode mailing list