On Wed, 22 May 2013 23:09:54 +0200 Irek Szczesniak wrote:
> On Fri, May 17, 2013 at 4:43 PM, Irek Szczesniak <[email protected]> 
> wrote:
> > On Fri, May 17, 2013 at 3:07 PM, David Korn <[email protected]> wrote:
> >> Subject: Re: Re: [uwin-users] uwin-users Digest, Vol 98, Issue 5
> >> --------
> >>
> >>> my guess is the uwin fork/exec code would run afoul of wine's emulation
> >>> that itself probably uses fork/exec and stands on its head trying to hide 
> >>> that
> >>
> >> I don't know why this would be a problem assuming that the WINE
> >> emulation of CreateProcess() doesn't randomly lay out the address space.
> >> That means that two calls to CreateProcess() with the same arguments
> >> should put shared libraries at the same address.
> >>
> >> UWIN uses only WIN32 calls so that if WINE implements WIN32 faithfully,
> >> then in theory UWIN would work on WINE.
> >
> > Does posix.dll implement posix_spawn()? IMO a native posix_spawn()
> > implementation would avoid the trouble of a fork(), exec() sequence

> Glenn?

posix.dll provides spawnveg() with similar but limited semantics to 
posix_spawn()
libast spawnvex() uses posix_spawn() or spawnveg() if possible
however fork() is required by ksh -- well ksh can run on non-fork systems but 
its
not nearly as efficient -- dgk can provide more details on that

_______________________________________________
uwin-users mailing list
[email protected]
http://lists.research.att.com/mailman/listinfo/uwin-users

Reply via email to