Re: Mint library ramblings...

[Stephen Usher <steve@earth.ox.ac.uk>]

>For testing I was compiling Eric's port of the GNU fileutils.  I
>decided to try to figure out why cross-device mv's have never worked
>correctly on my system.  It seems that the copy() routine it uses to
>copy from one filesystem to the other does read()s and write()s with a
>size of 8K, and this is what was causing it to have strange results.
>
>Is it a known bug in earlier versions of GEMDOS that you can't
>reliably Fwrite() more than about 1K at a time?  If so has it been
>fixed?  Or is the problem elsewhere (ICD driver?  I think I had this
>same problem under AHDI 3.01 as well.)

Hmm.. well a copy of cp I wrote which tried to copy as much as possible of a
file at a time always worked under both MiNT and TOS (1.02, 1.04 and 3.05).
I've used it under AHDI 4.xx and 5.xx and with the ICD utils from version 3
onwards.. maybe there's still a bug in the fileutils mv across filesystems.

>The consequences of these long writes can be anything from an address
>error to a total system crash (usually with a spectacular amount of
>trash dumped to the console).
>
>Another unrelated problem (feature?) that I noticed is the library's
>behavior of chown() on TOS partitions.  Currently it gives a -1
>return, with errno indicating an invalid function.  My feeling is that
>this is a bit too paranoid, and that it would be reasonable to return
>success from chown() if Fchown() returns EINVFN.  Or is this just
>asking for trouble?

I think it, and chmod(), should return success else some installation
programs and makefiles will break needlessly. Just because this filesystem
doesn't support the notion on ownership or the execute bit doesn't mean that
the program which has tried to do the operation is in error, after all, the
files seem to be owned by the user trying to do the operation, how on earth
is the program going to know that it can't work?

>
>cheers,
>entropy
>

Steve

-- 
---------------------------------------------------------------------------
Computer Systems Administrator, Dept. of Earth Sciences, Oxford University.
E-Mail: steve@uk.ac.ox.earth (JANET) steve@earth.ox.ac.uk (Internet).
Tel:- Oxford (0865) 282110 (UK) or +44 865 282110 (International).