Re[2]: Experimental UNICODE-only MinGW Build]

[Artem Gr <artem at bizlink dot ru>]

I am more than confused already.
It seems there is some different patches,
all of them ready to be checked in,
(although I saw only the
http://gcc.gnu.org/ml/java/2003-11/msg00258.html ).

Excuse me for interfering without having a much knowledge on the
subject, but some questions are tempting to be asked here.

ME> then I will add the configure switch
ME> and test both configurations exactly once.

Why a switch is required? Or even an extra #define?
Aren't we aware already about an OS being the target?

Secondly, aren't we too eager to fix Win9X problem now?
Isn't it better to use the minimalistic, "small steps", approach:
fix the problem for UNIX and WinNT, make a test case,
and than think about going further?

Thirdly, is it possible to hide a conversions behind an interface, so
that libjava is all UNICODE and all conversions are encapsulated
somewhere? The patch I saw so far is rather dispersed.

Finally, and least important: it is my personal opinion,
but for my taste the inclusion of libunicows is an overkill
for a "fast patch". Such modifications are better left for later,
when there is a time to carefully desing and balance everything.

Artem