Dxreaddir/Fsfirst-bug

[Holger Kleinschmidt <holgerkl@cs.tu-berlin.de>]

Michael Hohmuth wrote:

> Holger Kleinschmidt wrote:
>
> > With most (but not all) directories the modification times of the
> > first two entries ('.' and '..') are wrong, althoug Fxattr gives the correct
> > ones. The directories I've tested reside on the root of a TOS-filesystem,
> > e.g. 'f:\cat'. Fxattr gives the correct date of creation for '.' and Zero
> > for '..', but Dxreaddir gives the same wrong date for both entries.
> > For one directory (out of six) on the same drive Dxreaddir gives the correct
> > results, but I can't see any reason why this directory should be special
> > (it's not the first or last one).
>
> Tosfs uses the underlying TOS' Fsfirst()/Fsnext() system calls
> (Fxattr(dir/.) uses Fsfirst(dir/.), Dopendir(dir)/Dxreaddir() use
> Fsfirst(dir/*.*)).  From a first look, I'd guess that you've found a
> bug in your TOS version and not in MiNT...

It seems you're right: I've tried again with plain TOS 2.06, and GNU-ls
as well as Gemini/Mupfel-ls show the wrong date for '.' and '..'.
I havn't been aware of this before. Is this only a "feature" of 2.06 or
do other versions have the same bug?

Bye, Holger