RE: [MiNT] /proc

["Julian Reschke" <reschke@medicaldataservice.de>]

> From: owner-mint@gfanrend.fishpool.fi
> [mailto:owner-mint@gfanrend.fishpool.fi]On Behalf Of Thomas Binder
> Sent: Sunday, November 14, 1999 5:11 PM
> To: mint@gfanrend.fishpool.fi
> Subject: Re: [MiNT] /proc
>
>
> Hi!
>
> On Thu, Nov 11, 1999 at 09:54:00PM +0100, Jörg Westheide wrote:
> > The only real solution for that is to make it configurable.
> > N.AES uses the environment variables ACCEXT, GEMEXT and TOSEXT for that
>
> But this can't be used inside procfs, as - at least IMO - it's not the
> kernel's task to evaluate environment variables, especially not for
> something which is not specific to one process. For the same reason, my
> implementation of MagiC's Slbopen() does not evaluate $SLBPATH when
> called with a NULL pointer for the path.

Conceptually you are right, but I think there is a very good reason why
Andreas has implemented it this way: otherwise, the minimal code to load a
shared library is much bigger, because the (statically bound code in the)
client needs to do the path search on his own.

Could you be a bit more specific in which ways your implementation differs
from MagiC's and/or from the one I have specified? Maybe it would be a good
compromise to allow to pass a path list instead of a path, and have
Slbopen() do the path search then (without having to access the
environment).

Regards, Julian