Re: [win32] -Djava.library.path

[Marco Trudel <mtrudel at gmx dot ch>]

Tom Tromey wrote:
> > > > > > "Marco" == Marco Trudel <mtrudel@gmx.ch> writes:
>
> Marco> I use GCJ 4.2.0 (20060630) on linux and windows.
> Marco> While passing -Djava.library.path=foobar to GCJ leads to the expected
> Marco> behaviour on linux, it seems to be ignored on windows.
>
> Marco> Is this a known bug/problem (I didn't find anything on google or the
> Marco> GCC bugzilla) or shall I make a bug report?
> Marco> Has anyone an idea for a workaround?
>
> Offhand I would expect this to work ok.  You might verify that
> USE_LTDL is set on Windows; see the two definitions of _Jv_SetDLLSearchPath
> in natSystemProperties.cc.

I will do that the next time I compile GCJ for windows. I'll come back 
to you.


> Filing a bug report is a good idea.  Put '[win32]' in the synopsis.
> Note though that Windows-specific things are worked on very rarely; in
> fact we're having a problem getting an existing patch tested on
> Windows, and this is turning into a blocker.

I know but don't really understand. I mean, for most of you guys, this 
really is easier than to tie your shoes...
Is it a kind of "I hate windows" thing, or is it just too painful to 
crosscompile GCJ and test it on windows (with vmware or similar, linux 
doesn't even have to be shut down ;-)).
But ok, we could arrange that I could test stuff regularly. Lets say 
once a week or once every two week... Interested?


thanks
Marco