On Mon, 24 Feb 2003, Jon Hylands wrote:

> At Monday, 24 February 2003, you wrote:
> 
> You know, this is starting to spin out of control.

W.r.t. Unicode in the image, I agree completely.  The required changes
would not be trivial.

> This started out as simply taking the Apple fonts out of the image,
> and now we're talking about replacing the entire character system.

I think the discussion about Latin-1[5] is entirely reasonable.  It
doesn't alter the model within the image at all (only the correspondance
between chars > 127 and the symbols that are displayed for them).  
Adopting the ISO encodings eliminates other (more painful) problems that
would present themselves if we were to attempt a "halfway" solution (such
as replacing the fonts but retaining MacRoman encodings).

> Remember, small steps, simple increments, etc...

Replacing the Mac fonts with Latin fonts and then modifying (very
slightly) the way the various VMs handle conversion during import/export
of text really is the fastest and most reliable route to having a new set
of fonts that work consistently -- and represents the least amount of
work.

Since this would introduce incompatibilities with accented characters,
ligatures and symbols, it would certainly be made as part of a version
increment.  The 3.4-series images and VMs would remain as they are today,
and (for example) the 3.5 series would move to Latin-based fonts and
encodings. The 3.5 VMs would be backwards compatible with earlier images
except where accents, ligatures and special symbols are concerned.  
Conversion to/from the majority of MacRoman to the majority of Latin-1[5]
is trivial and can be done by table lookup.  So it would be very little
work to add (for example) a command key to the ParagraphEditor in 3.5 to
convert the old MacRoman encodings to/from the new Latin encodings
(similarly to the existing key that converts CR line endings to/from LF).

Ian