Re: [MiNT] STAT bits - Ouch!

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

On Fri, 2010-12-10 at 23:25 -0500, Roger Burrows wrote:
> Hi all,
> On 10 Dec 2010 at 22:03, m0n0 wrote:
> > >
> > > You didn't say this, but I did. If POSIX doesn't define them, then, as you
> > > yourself noticed in your first post, MiNT is in the very position to
> > > define them arbitrarily and POSIX-compliant programs have to work. If they
> > > do not, then they are not POSIX-compliant, kinda by definition.
> > 
> > I fully agree to the this. The applications should be changed, not the OS.
> 
> This is a counsel of perfection, alas. If you have non-conforming applications, 
> the options are:
> 1. Don't run the application.
> 2. Change the application.
> 3. Change the OS.
> 
> I don't think that anyone here would suggest that #3 is the best choice if #2 
> is feasible.  However, I understand Alan to say that the maintainers of git & 
> python are unwilling to change their applications.  For some reason, the fact 
> that their code doesn't work on FreeMiNT doesn't worry them :-(.  So if we want 
> to run them on FreeMiNT we either patch the applications every time there is a 
> change (extra work for python and difficult/impossible for git), or we change 
> the OS.  Doing the latter will also help if there are other, as-yet-
> unidentified programs which rely in the Un*x definitions.

I didn't say they were unwilling. I've never asked the Python folks,
only the git folks, and that was non-trivial.

They did ask what OS differed, and I mentioned FreeMiNT. So, I'm not
sure how bothered they are.

My whole point of the initial email was to identify the options of 2 and
3 above. If we don't change we need resources to fix the applications.
I'm not sure how abundant they are.

Alan.

Alan.