Re: Availability of various calls

[mk@anuurn.do.open.de (Martin Koehling)]

In <9508171054.AA02040@mail.imc.exec.nhs.uk>,
Chris Ridd (chris@imc.exec.nhs.uk) wrote:

>Martin Koehling wrote:
>
>>MiNT implements Dopendir/Dreaddir for all drives, but MetaDOS
>>doesn't - apparently there are MetaDOS drivers out there that
>>implement these functions only for CD-ROMs (in order to provide
>>access to CD-ROMs using long filenames)...
>>
>>The best (and safest) way to handle this appears to be to *always*
>>call Dopendir() first, and only to use Fsfirst() if Dopendir()
>>failed with EINVF (-32L)...
>
>OK, I'll need to have a flag for each available drive, in that case. With
>some fudging for accesses to U:\...

What do you need these flags for?
Simply *always* calling Dopendir() and checking it's return value
is both fast and simple (and thus *reliable*)...
And besides: what would you do about file systems that *don't*
reside on one of the usual BIOS drives (0..31)?

>>I think it's safer not to depend on this - a future version of
>>MagiC (the operating system) might implement only a subset of MiNT's
>>signal handling facilities; I'd better always check for EINVF...
>
>Yes, I suspected this could happen. But I can't decide if implementing
>Psigaction makes sense without implementing Psigblock, for instance.
>
>Do I take it then that the current release of MagiC!3 doesn't support
>signals?

Yep - it doesn't. :-(
It *does* implement many of the new MiNT calls (in particular
Dopendir/D(x)readdir/Fxattr), drive U: and pipes (which are
necessary for the drag&drop protocol), but it still lacks many
of the more Unix-like MiNT functions...

Martin