RE: [MiNT] /proc

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

> On Wed, Nov 17, 1999 at 04:28:28PM +0100, Julian Reschke wrote:
> > > Of Konrad M. Kokoszkiewicz
> > > I meant, that the kernel should "accept SLBPATH"... if this variable
> > > points to just one folder (SLBPATH=u:\c\multitos\slb), the
> kernel could
> > > just paste it into the library pathname, so that no disk searching is
> > > needed to be done. But well, I don't know if Gryf likes that
> idea, because
> > > it again smells like evaluation of the variable :-)
> >
> > Sure, it would be nice if path searching could be kept outside
> the kernel.
> > However the alternative is that each client will have to do it
> instead --
> > that doesn't sound much better...
>
> I thought the name "SLB" had something to do with shared libs?  Usually
> the dynamic linker is a shared lib itself.  Why don't you just put the
> actual loading with the path evaluation into a shared object?  Peeking
> into the process' environment is a little too much magic for me...
>
> Maybe a stupid answer, I have never really looked at the slb specs.

Then do it :-) The SLB scheme currently doesn't have a separate loader. The
programs are responsible for loading the SLBs on their own.