On 12/12/2001 09:28:43 AM Marco Cimarosti wrote:
>What you cannot easily do with Keyman is visual order input - E.g. short
i
>before consonant, repha ra after consonant.
Have you given it a try? Once you figure out what it means to have a short
i with no consonant and what you want it to look like in that temporary
state, you could probably get it to do it. I have used Keyman to create
some input methods that did some similar things (though not for Devanagari
script).
>Moreover, it is tricly to implement the deletion of the <danda>: when the
>user hits backspace near <0917> the keyboard script should rather add
<094D,
>200D>.
That's easier than the short i input problem -- it's quite easy, in fact.
The only limitation is that Keyman input methods are only aware of what is
in the text context as long as you keep entering keystrokes without any
intervening mouse clicks, arrow keys or change of focus. This will change
once it supports Microsoft's Text Services framework, and as apps also
support that (Office does now). That will allow Keyman to always be aware
of what is in the text context, even when the user mouse clicks into
existing text. This functionality is coming in the next version.
>Finally, there is no way to simulate placing the cursor between the <half
>ga> and the <danda> glyphs.
Well, Graphite could make it possible to create an implementation in which
you could select the left half of the ga independently from the right
half, except that there wouldn't be any character in the backing store
that that could correspond to. That's a problem even in your proposed way
of dealing with these things, isn't it? What this would require is for the
rendering system to be able to report back to the app some selection range
which the app could pass on to the input method -- and there would have to
be some way to express that selection range that the input method could
understand. Then the input method could recognise the state and have
appropriate actions; e.g. if the user hits [Delete] it would insert <
094D, 200D >.
Of course, we need to recognise that where there is this mismatch between
the logical units that are presented to the user and the logical units
that are actually in the backing store, we are inviting usability
problems, no matter how much cleverness we try to put in.
- Peter
---------------------------------------------------------------------------
Peter Constable
Non-Roman Script Initiative, SIL International
7500 W. Camp Wisdom Rd., Dallas, TX 75236, USA
Tel: +1 972 708 7485
E-mail: <peter_constable@sil.org>
This archive was generated by hypermail 2.1.2 : Thu Dec 13 2001 - 00:53:04 EST