Re: MiNT goes UNiX, invitation for mailing list (MINTOS)

[Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>]

Chris Herborth <herborth@53iss6.waterloo.ncr.com> writes:

|> What you wrote:
|>> 
|>> walra%moacs11@relay.NL.net (Waldi Ravens) writes:
|>> 
|>> |> Today I looked at a few socalled ports, and noticed that the atari switch
|>> |> was mainly used to:
|>> [...]
|>> |> 2. compromise to the Gemdos filesystem limitations
|>> |>    not necessary for the minix fs, and theoretically there could be a
|>> |>    gemdos.xfs to overcome the gemdos fs limitations
|>> 
|>> In fact, i have written such a thing, based on _unx2dos in the
|>> GNUlibs.  I have once planned to distribute it, but since we have
|>> MinixFS, there is no need any more, i think.  Anyway, i'm still using
|>> it as a replacement for tosfs in MiNT.

|> Does this emulate the Minix fs under a TOS filesystem?  That'd be great!
|> Is it as compatible as the minixfs (ie, it gives unique 8.3 filenames to
|> brain-dead programs so you can still access your files)?

Well, it emulates as much as _unx2dos from GNUlib emulates, i.e.
symbolic links and up to 32 character file names. It uses the same
mapping from long to 8.3 file names, equivalent to UNIXMODE="LAH.,"
(but it isn't configurable, only by recompiling it).  A program
running in the TOS domain only sees the mangled 8.3 file names and no
symbolic links, the names are simply untranslated in this case. (In
most cases, the mangling just means truncation, but the names are
guaranteed to be unique, of course.)

Unfortunately, it is even slower than tosfs, and since we have
MinixFS, i would not recommend it to anyone.

----------------------------------------------------------------------
Andreas Schwab                                  "And now for something
schwab@ls5.informatik.uni-dortmund.de            completely different"