[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [MiNT] MiNTLib 0.52.3b

> From: Frank Naumann [mailto:fnaumann@prinz-atm.CS.Uni-Magdeburg.De]
> Sent: Friday, August 06, 1999 2:27 PM
> To: Julian Reschke
> Cc: mint@fishpool.com
> Subject: RE: [MiNT] MiNTLib 0.52.3b
> Hi Julian!
> > > As you said this is moving and not renaming! Moving across
> filesystems is
> > > another thing.
> >
> > As it was said before, moving and renaming is the same thing in
> POSIX speak.
> I think we speak here about filesystems?
> > > Which 'normal' programs are interested how many different
> filesystem do
> > > you have?
> >
> > df.ttp, du.ttp, ...
> Right, such system tool must be updated.
> > backup programs, program installers, desktops,
> And that's the problem to go through u:? And if a filesystem is mounted
> under /f/tmp it's not the recommended way that it's only accessible
> through this point?

There is always a need to programmatically get a list of filesystems. In the
past, this was by using the drive bits. In fact, this was not really a clean
way, because there might be GEMDOS devices which are not BIOS devices, so
programs hopefully did a Dsetdrv (Dgetdrv()) instead.

This in turn was broken by MiNT, where filesystems can appear in U:\ without
havinf *any* GEMDOS drive letter. So currently properly written programs
*must* use the entries in U:\.

I still don't see any *convincing* reason why this has to change. If you
want to mount filesystems somewhere else, go ahead and do it, but keep the
entry in U:\ and hide it. Shells and desktops by default will not display
it, and everything is fine.

The concept is OK, don't break it.