Re: quail inserts raw characters on unfinished sequences

[Juri Linkov]

>> This could be fixed, when you agree with the change.
>
> Ok, I know agree with the change.  Please install your
> change after fixing it.

I fixed and installed it.

> But, please suppress this facility for such
> non-deterministic input methods as chinese-py.  You can
> check it by `quail-deterministic'.  I believe it's too
> annoying to see intermediate partial translation in such
> cases (e.g. you'll see what I meen when you type "chua" in
> chinese-py).

I see.  There is no way to select a different alternative
for the last untranslated character.  I don't know if it is useful
to implement such selection for non-deterministic input methods
(finishing the input sequence just after "chua" leaves in the buffer
one character translated from "chu" and latin "a", but maybe
more preferable would be two characters: one character translated
from "chu" and another character translated from "a", but I can't
argue about non-deterministic input methods), so I suppressed
non-deterministic input methods now.

> To input "шч", you must type, for instance, "sh\C-\\C-\ch".
> If the input method is unambiguous, one should be able to
> input any character sequence without such a trick.

Actually there is no such problem for Cyrillic-based languages like
Russian and Ukrainian where words with "шч" in them are very rare.
But other languages like Belarusian have many words with "шч".  OTOH,
other rules in cyrillic-translit are ambiguous for Belarusian too,
so this input method doesn't work for it, and users of such languages
should use other input methods.

-- 
Juri Linkov
http://www.jurta.org/emacs/