Mac OS X Problems (character encoding)

[Joel Rees]

I have to preface this with a disclaimer -- I am NOT criticizing Scott
or any of the technical teams at runrev/mc. I am not really criticizing
the Unicode Consortium, either. I am criticizing the entire industry. So
make sure any flames are aimed correctly ...

And Scott replied, 
> On Tue, 08 Jul 2003 Richard Gaskin <ambassador at fourthworld.com> wrote:
> 
> > >> Yves wants to create new folders in the Finder. A custom font would
> > >> require that he set the Finder to use that font, which could
> > >> potentially alter any existing folder names containing diatricitals.
> > >> If Yves is planning to distribute his stack to others, it would
> > >> require that all his users also install the font and risk the same
> > >> changes to their existing folder names. I don't see it as a viable
> > >> solution.
> > > 
> > > I think the only good solution is that the Rev team makes it possible
> > > again as it was with Rev 1.1.1
> > 
> > I guess the obvious next question is:  what is different between the 1.1.1
> > implementation and v2.0, and what went into the decision to change it?
> 
> The difference is that in 2.0 the engine was converted to Mach-O
> format and uses UNIX system calls to access the filesystem whereas 1.X
> used the older (and slower and much less capable) File Manager calls.
> The real problem here is rooted in the circa 1982 decision by the
> Apple developers to use a proprietary character encoding for the
> Lisa/Mac rather than the ISO standard. 

Scott, there were no standards then. Suggested standards, yes, and a
number of them. Going off on your own was in fact the correct thing to
do back then.

> This decision has now bitten
> them (and you!) in the ass because the display system uses a different
> character encoding than the UNIX filesystem does, making stuff like
> this very hard to deal with because you have to do character-set
> conversions all over the place.

That is entirely misplacing the blame. It's not Apple's fault, any more
than it's Metacard or RunRev's fault. 

> It should be possible for us to fix or at least work around this in
> the engine, but in the mean time using a script-level workaround is
> your best bet.

There is no fix except for work-arounds. Unicode is huge, and it is not
really compatible with any other existing encoding. (Round tripping is
not the same as one-to-one correspondence.) It is still under
development in some of the most telling areas (example, rules for
collating), and that is where much of the pain comes from.

Pardon me for tossing out potential flame bait in my first post.

Joel Rees