Re: [MiNT] [PATCH] Machine dependent module path extension

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

On 02/13/13 13:36, Helmut Karlowski wrote:
> Alan Hourihane, 13.02.2013 13:24:43:
>
> > > > Then I don't understand why the filename contains the version too. It
> > > > makes a mess of the xaaes-folder every time a new minor version of 
> > > > XaAES
> > > > is installed.
> > >
> > > What mess?
> >
> > I guess you end up with....
> >
> > xaaes014.rsc
> > xaaes015.rsc
> > xaaes016.rsc
>
> Why? The rsc-version does not necessarily change when the 
> XaAES-version changes.

Maybe not, but the potential is there.

> > etc. But the XaAES binary is a single file, and you don't know which 
> > RSC file it uses. Obviously you assume xaaes016.rsc as you think 
> > you've got the latest XaAES binaries. But you no longer need 
> > xaaes015.rsc and xaaes014.rsc, but how would you know ??
> >
> > We don't have things like ext2fs01.xfs, ext2fs02.xfs etc, so I guess 
> > why should RSC's be versioned in the filename when things are 
> > versioned and checked in the file themselves ??
>
> Because you know IN ADVANCE which rsc-version you have and don't have 
> to start XaAES to check it, and XaAES knows which to load in case 
> there are different versions in parallel for testing etc. These 
> numbers only grow, so it's easy to determine those that can be deleted.
>
> There is only one version in the CVS.

You know as you wrote the code. It's less obvious for those that didn't. 
Plus why fallback to xaaes.rsc at all ?

I'm just saying that other files don't have versions in them.

Alan.