Re: Experimental UNICODE-only MinGW Build

[João Garcia <jgarcia at uk2 dot net>]

Hi Mohan,

Mohan Embar wrote:

> I guess the above statement isn't true anymore....:
>
> http://apnews.excite.com/article/20040113/D801L4C00.html
Riddles from the Dark... ;-)

Anyway... it's a mater of time. I guess it does not change anything 
about the ongoing gcj strategy.

For me, what is annoying in mingw-gcj at present time is that it seems 
to work better in the win9x branch... :-/

Threads and exceptions seem to be falling apart in the NT branch, 
specially if one is to use CNI (obviously, the cause can probably be 
reduced to the exceptions problem...).
There are things that work fine both in JVM (using JNI; in both platform
branches) and with mingw-gcj (using both JNI or CNI) in win9x 
environment but do not work at all with the same mingw-gcj in winXP...
And this is for the same source code, and happens both for gcj 3.4
and for gcj 3.3 (but not exactly in the same code segments)... Although
gcj 3.4 seems to be a little better.

I hope that these problems can be solved in time for the 3.4 release...
But I have no idea of what is necessary to solve them (to me, that area 
is a black box -- I am just using try/catch where it shouldn't be 
necessary just to make things work).


João