RE: VFAT.XFS versus MagiC

[Sven Karlsson <sven@it.lth.se>]

> If you want just the long filenames, an utility for reading&writting them
> under TOS/MiNT can be written in a hour or so. It's very easy, believe it
> or not.

If I can get 2 hours this evening a simple VFAT reader that can read any
FAT12 partition (ie floppies) with VFAT is done. (It currently can read
VFAT names and traverse the directory tree.) All that's needed is a 
very simple file copy routine.

> Remember, that without this driver the VFAT or VFAT32 support is useless.
> And the other way round: someone will write a VFAT(32) filesystem when the
> big partitions can be read under MiNT, because then we will really need
> it. Writting XFS for diskettes it's really a wasting of time.

I agree. What's really needed is a FAT(whatever) system that replaces TOS'
routs completely. I'm writing such a filesystem for Fenix which also have
support for hardlinks, mountingpoints and other unixfs features not preset
in the normal DOS/TOS filesystems. It also have a more "unix fsish" block
allocation routine so that fragmentation is avoided if possible. However
this is for Fenix which have a slightly altered fs binding. I know my
commitment to Fenix might disturb you all but I have to choose between
supporting MiNT and try to dodge all bugs and add kludges to a mega patched
system or make a fresh start with a (as far as possible) compatible system
which can replace MiNT AND TOS. Like all longterm software projects needs
rewrites sometimes I really think MiNT needed one around 92 so it is really
getting too late. :-( A rewrite is painful but needed to let a system evolve
without too much maintaining effort.

What's the bottom line?

1) During a hopefully very short period I (and a couple more programmers) can
   offer limited support to MiNT.
2) After this period there will be a painful period of betatesting where a lot
   of bugs will show up. During this period some end users can use Fenix instead
   of MiNT/TOS if they are brave enough. Finally Fenix will be a rather stable
   system that everyone can use and will also run native on PPC or Intel systems.
   There will probably be bugs (like in all software) but since all source code
   to Fenix will be freely obtainable patches can be made quickly.
3) To the end user this means that MiNT might not evolve for about 6 months. However
   small patches and workarounds can be made. (Like a VFAT <-> minixfs conversion
   utility).

This does not mean that MiNT is a bad system at all! It works perfectly well as a
Unix and most unix tools can be compiled for MiNT without any problem. Fenix is
based on MiNT and is basically a cleanuped MiNT with BIOS,XBIOS and non MiNT TOS
routs thus eliminating the need to run Fenix on top of TOS. This have several
advantages one being removing the annoying TOS workarounds that slows down serial
communication.

As I said I can offer limited MiNT support/commitment until Fenix becomes usable for
the end user. I therefore suggest that we should discuss what features that are most
needed in MiNT (like VFAT) and make a priority list.

Also if you feel like you would like to help/work within the Fenix project just mail
me and I will get back you you. It doesn't matter if you are a graphics/music artist
or a Linux wizard. Help is always needed!

Regards
 Sven