Re: [MiNT] STAT bits - Ouch!

[Alan Hourihane <alanh@fairlite.co.uk>]

On Thu, 2010-12-09 at 00:52 +0100, Konrad Kokoszkiewicz wrote:
> > Oh, and the other nice thing is this in our own code for ext2....
> >
> > # if EXT2_IFSOCK != S_IFSOCK
> >                if (EXT2_ISSOCK (mode))
> >                {
> >                        ptr->mode &= ~EXT2_IFSOCK;
> >                        ptr->mode |= S_IFSOCK;
> >                }
> > # endif
> >
> > The above is because if you bring in a filesystem from another Linux
> > machine that uses EXT2 we have to do conversion. Yuck.
> >
> > It makes sense to move to this "defacto" standard.
> 
> No. ext2.xfs is a device driver, subordinate to the kernel. It has to do 
> as many data conversions as it is necessary for the kernel and (if the 
> data reaches the apps) applications.

It's actually a performance enhancement by "not" having to do this.

> I would be against the change until more evidence is available. For now, 
> only two programs have been adduced, one being a programming language (and 
> the problematic file is *.py), and nothing has been quoted from the other 
> one (git? or what), that would make it evident that MiNT's stat() is to be 
> changed.

I can understand the desire to stay with what's current, hence my email
to voice opinions.

I'll leave it for now, given yes, it's just python and git and gather
more evidence as I go. But at least the problem is out in the open for
others to bump into potential applications that can cause issues here.

Alan.