Re: Filenames with accented characters

[Mohan Embar <gnustuff at thisiscool dot com>]

Hi João,

>It is also my objective in this post to get some light over certain 
>things... If someone is willing to give me such a light (even if it is 
>in solid form ;-)  -- it would be better than simply getting no reply at 
>all about those things)...

...

>I CAN NOT understand why something that can be so simple gets so 
>complicated...

Like Ranjit, I too didn't take anything you said as a personal attack.
On the contrary, I appreciate the knowledge you have on this subject
and your efforts to remedy it.

I've lived in France for several years and am aware of issues of
linguistic discrimination in software. So rest assured that I take this
problem seriously. There are several issues, though, that have
made me less motivated to solve this problem than other problems.

First and foremost, I don't understand enough about character
encoding and conversion as I'd like, though Ranjit's and your
posts and patches have certainly helped. I still feel like I need
to do more studying before I feel comfortable, though.

Secondly, the issue of internationalization within libgcj is just part of
the entire solution. Currently, since mingw32-gcj is compiled without national
language support, my compile chokes even when I have accented
characters in the source. Danny Smith philosophized over the inclusion
of libiconv in mingw32-gcc, but as far as I know, we haven't gotten
to the root of this problem yet:

http://sourceforge.net/mailarchive/message.php?msg_id=5704041

If libiconv were part of the solution, this could potentially change
our implementation.

Thirdly, there is the issue of Win9X support. Since:

- not supporting Win9X would make our solution more elegant,
- I feel we should support Win9X anyway so as not to alienate users
- I personally don't much care for Win9X

...this has added to my procrastination.

Fourthly, there is my desire to do this properly and for all targets, as with
this post:

http://gcc.gnu.org/ml/java-patches/2003-q3/msg00697.html

And finally, I had minor issues with your patches. As far as I recall, they
were 3.3-specific only and had too much code or code duplication for
my taste. I also recall an initial patch which you said wasn't 100% correct
and would need further troubleshooting. As with others, the easier it is
for me to incorporate and verify something, the more likely it is that
I'll give it serious attention. Only recently, you suggested a solution
for 3.4 which, although it still duplicates an entire class and macro in jvm.h,
involves far less duplication than your previous patches. The problem,
though, is that the email was in the form of suggestions and not
a real, tested patch which I can apply with my eyes closed. And here again,
I am reluctant to promote this as the "final solution" before we can
investigate Bryce's suggestions and do this properly.

>But Mohan has the "Binary Power" in this matter as far as I know, so 
>it's Mohan's rules... ;-)

Like I've said on several occasions, I am in no way averse to receiving
a patch from you which applies cleanly to the codebase of your choice
and pushing out a build on my website called "João's Build". Therefore,
you share in this awe-inspiring binary power that I have.

Thanks again for all of the knowledge and insight you're provided me
concerning this matter, and sorry that things aren't moving quickly enough.

-- Mohan
http://www.thisiscool.com/
http://www.animalsong.org/