[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Wanted: Superstar hacker to port Mono to hurd-i386

On Thu, 2012-02-02 at 12:28 +0100, Pino Toscano wrote:
> Hi Jo,
> Alle giovedì 2 febbraio 2012, Jo Shields ha scritto:
> > Mono is supported on most Debian architectures. Including
> > kfreebsd-i386 and kfreebsd-amd64. Know what'd be nice? Support for
> > hurd-i386. It'd bring in a bunch of nice user-friendly apps to
> > hurd-i386, which would be good.
> > 
> > http://www.mono-project.com/Porting#Operating_System_Ports notes that
> > OS ports aren't too much work if the OS is POSIX and the CPU work
> > has already been done, so I don't think this would be an outlandish
> > amount of work for any skilled Hurd hacker.
> > 
> > Any takers? I'd really like to see this running on all the Debian
> > architectures, if possible.
> A couple of weeks before the mono transition, I had a try in building 
> the (formerly in experimental) current mono in unstable.
> Basically, the chanages I needed were about:
> - fixing the embedded libgc copy like it has been done upstream or in
>   gcc (it was mostly cherrypicking the changes)
> - adding hurd-specific bits (like in configure.in, mono-config.c,
>   mini-x86.h)
> - fixing "#ifdef defined(__MACH__)" to check for both __MACH__ and
>   __APPLE__, since those bits are MacOSX-specific and __MACH__ is
>   defined for us too (since our microkernel, GNU Mach, is derived from
>   Mach)
> - the usual PATH_MAX/MAXPATHLEN lack (I put #ifndef ... #define to just
>   have a version quickly compilable)
> and then I got stuck at the .dll generation phase, because of mcs.exe 
> going on starvation, with one mcs' thread waiting on a mutex lock (IIRC) 
> and one libgc's thread on a sem_wait.
> Few days later, trying to compile a different source, ecl, gave a very 
> similar starvation issue, so most probably there's some bug on our side 
> we need to fix before getting mono compiled on Hurd (but I didn't 
> investigate neither mono nor ecl that much regarding to this).

Has your almost-works-but-not-quite-damnit port been committed anywhere
public? Since it seems you already did 95% of the work, it'd be a shame
to let it go to waste, especially is Harish is also interested in
getting this working

Attachment: signature.asc
Description: This is a digitally signed message part

Reply to: