[BusyBox] Re: Fix for "more" and "ps" on dumb terminals

Erik Andersen andersen at lineo.com
Fri Jun 2 17:36:11 UTC 2000


On Fri Jun 02, 2000 at 09:15:06AM -0400, Pavel Roskin wrote:
> Hello, Erik!
> 
> First of all, CVS is basically Ok. There is one minor issue. I cannot see
> available modules by running "cvs co -c". You should check out CVSROOT and
> add to the file "modules" at least following:
> 
> busybox busybox
> tinylogin tinylogin
> 
> And then check "modules" in. Now check what "cvs co -c" shows.

oops. Fixed now I believe.  Thanks for the tip.

    [andersen at dillweed busybox]$ cvs co -c
    andersen at busybox.net's password:
    busybox      busybox
    modules      CVSROOT modules
    tinylogin    tinylogin
    uC-libc      uC-libc

> > > By the way, you may want to put some common code in both versions of
> > > ps_main() to a separate function if you happen to hack ps.c
> > 
> > Yeah.  I've actually been debating removing the devps patch stuff,
> > since Linus has given the decree that it will never go into the kernel.
> > What do you think?  Should I axe it?
> 
> I think you shouldn't remove your patch. The reason are following:
> 
> 1) There are systems that are really short in non-volatile memory.
> Actually, the system that my company produces, has just 4 Mb of onboard
> flash. Fortunately we could make 32 Mb ATA Flash cards work, but we had a
> backup plan. devps would be the way to go.
> 
> 2) Linus doesn't like your patch today, but he can change his mind during
> the 2.5 series.
> 
> 3) Busybox is basically intended to be run on systems with kernels built
> for those systems, not with prebuilt kernels from e.g. RedHat or
> YellowDog. Hence patching the kernel is not a problem.
> 
> 4) GNU Hurd can only implement mountable filesystems, such as /proc using
> translators, i.e. userspace programs. The translator should have another
> way to get information about processes from the microkernel. This means
> that devps (in some form) is not only easier to implement in GNU Hurd, but
> that it's probably required to implement procfs. The same may apply to
> other microkernel OS'es.
> 
> 5) As long as you keep the code duplication minimal it is not such a big
> deal to support devps.

Thanks for the vote of confidence.  I was kindof bummed when Linus
rejected it (I had put a fair bit of work into making it) and was
thinking my work had been wasted.  Thanks for the good reasons why 
it was worthwhile.  :) 

BTW, we need to get you CVS write access.  Want to use pserver or ssh?

 -Erik

--
Erik B. Andersen   email:  andersen at lineo.com
--This message was written using 73% post-consumer electrons--





More information about the busybox mailing list