Re: Thread deficiencies
On Mon, 3 Jan 2000, Alan Cox wrote:
>
> > Let me be more clear. I would like to suggest that the Linux threads be
> > "waivered" until POSIX threads are implemented and tested... The goal
>
> That seems foolish. They work.
I agree with Alan. But... what is the utility of
waivering Linux threads until POSIX are implemented?
It is just a matter to wait for a medium-short time.
>
> > GNU/Linux systems; however, what happens when Debian releases a GNU/Hurd
> > system? Also, Intel UNIX systems like SCO, Solaris, OpenBSD, FreeBSD, and
> > NetBSD have a GNU/Linux emulation mode, and it is possible to run NetScape
> > and WordPerfect with a proper sandbox of GNU (ie., LSB) libraries; however,
> > it would not be possible to run Linux threaded applications like
> > StarOffice. Pthreads provide the portability and runtime compatibility
> > between GNU systems with potentially different kernels.
>
> If they have a linux environment they support clone. If they dont support
> clone they are lying when they claim to have a linux compatible sandbox.
> Linux emulations on BSD get this right already - nobody else has any excuses.
The compatibility beetwen different kernel
for Unixes for Intel platform is a false problem.
Portability is a different matter. It is important, but we should not be
ossessed. it seems to me that sometimes a non portable solution
could be the best one, if you are sure you are right, then others will
come to you. maybe Alan is going to far talking
about lying, also if technically he is right.
The point is, if they want to be compatible, it is
their turn to implement compatibility and portability, if we
are respecting standards, then i do not see why cut off ourselves
for others.
>
> In addition packages already use the clone features, so we need to specify them.
> Trying to bury a Linux feature so you can implement poor compatibility with
> Linux is just crap.
>
> What exactly is your agenda for trying to kill off clone() ? Easy Linux
> emulation in AIX?
maybe easier emulation in IRIX ? (but who cares?)
>
> Alan
Luigi Genoni
Reply to: