[ut2004] No love with linux installer

Ian Hastie ianh at iahastie.clara.net
Sun Mar 21 08:18:11 EST 2004

On Saturday 20 Mar 2004 19:45, Jeff Craig wrote:
> On Sat, 2004-03-20 at 18:18, Sean Hamilton wrote:
> > Had these CDs been distributed with rockridge extensions, or
> > with 8.3 filenames, this wouldn't have been a problem. Both
> > of those are totally sane expectations. There is no practical
> > reason for the files on the CD to not be 8.3. Of course, these
> > are issues with the publisher, and I doubt they are open to
> > suggestion.
> 8.3 filenames are not necessary with modern filesystems, so it's not an
> unreasonable expectation that you be able handle extensions that allow
> for this.  Is Joliet the best extension?  Probably not, but as these
> discs were designed to be read under Windows as well, it's a lot easier
> to use Joliet from the publishers point of view.

Yes, as it will most likely be the option that requires least effort from the 
MS Windows users.  They do still make the vast majority of the software 
market, especially in PC games.

> > Who wrote this thing? Does he read this list? Can he comment?
> > Justify these decisions?
> The Loki_Installer


> I agree, theere are some serious design flaws inherent in that software.
> > Evidently your definition of ease is flawed. GTK interfaces
> > are no substitute for good design.
> The installer was designed with a few assumptions in mind.  In a modern
> Linux kernel, Joliet extensions are not wholly unreasonable, though
> there have been others who have had this issue.

This really doesn't make sense.  The installer should not need to know 
anything about the underlying filesystem.  It is not it's responsibility to 
try to do 8.3 to long file name conversion, that's what the Joliet extension 
option is for.  Any CD that must also be readable on a MS Windows system will 
be made using Joliet.  To expect the kernel to cover for an unconfigured 
kernel option is just unreasonable.


More information about the ut2004 mailing list