John Hudson wrote:
> Or prajnyenaatmanaasmaallokaadutkramyamusminsvarge !
OK, you win. I haven't a longer one.
> Marco, my comment was tangential to the overall discussion,
> and wasn't
> intended to divert your 'slow-motion' and very interesting analysis.
>
> >And Indian users too shouldn't be forced to process strings
> of *abstract*
> >characters into their heads!
>
> Agreed. But they should not necessarily be forced to process
> strings of glyph either.
Certainly not forced to. But what we see on the screen are glyphs, and many
people could find it more natural to interact with them, rather than
figuring how those signs map to codes inside the memory of the computer.
> The idea behind contemporary encoding and keyboard
> implementations for Indic scripts is that input should be
> phonetic.
Inputting new text is one thing, manipulating existing text is another. The
problem is the second thing.
> I don't think it is necessary at all to continuously think about the
> character/glyph model when typing. Most of the time, all you
> have to do is
> type the phonetic representation of the words, and let the
> shaping engine
> and font get on with the job of displaying it properly. It is
> only if you
> need to step back into what has already been rendered that
> you need to have
> some understanding of how that rendering has taken place.
This was my point: the problem is not typing new text, but moving through
the existing text to change something.
The sequence you typed (abstract characters) gets changed to something else
(glyphs), but this change is one-way only. Some of the keystrokes that you
type (namely the halant) do not correspond to manipulable objects on the
screen. So, once you type them, you cannot get back to change them.
Handled this way, the abstract characters become something similar to the
keystrokes of a CJK input method: you press many keys, and you get a single
blob. Once this blob (an ideograph) is on your screen, you can keep it or
delete it, but you have no way of going back and change it slightly.
If this may be acceptable and even natural for Chinese, it is a pity that
scripts having only 30 odd letters should work this way.
However, Dhrubajyoti Banerjee suggests that the hidden halants should become
visible (and hence manipulable) as soon as the cursor passes near them. With
this approach, part of my concerns above go away.
> The lesson: don't make mistakes when typing complex scripts :)
... Otherwise, you'll discover why they are called "complex scripts".
> Seriously, though, I am interested in your idea about a kind
> of editing
> level of rendering. This would be an implementation issue
> though, not an encoding issue, right?
Absolutely right. Maybe the whole discussion is even slightly off-topic, as
I am not discussing any enhancement or clarification to the encoding itself.
The only small concern with the encoding is the impossibility to represent a
stand-alone repha. But I guess that I have repeated this enough times now.
:-)
_ Marco
This archive was generated by hypermail 2.1.2 : Wed Nov 28 2001 - 13:20:24 EST