Adding Accufonts to the update stream (was Re: LicencesQuestion : Squeak-L Art 6.)

Ian Piumarta ian.piumarta at
Mon Feb 24 13:26:56 UTC 2003

Hi Cees,

On 24 Feb 2003, Cees de Groot wrote:
> On Mon, 2003-02-24 at 13:27, Ian Piumarta wrote:
> > this would be an opportunity to do away with MacRoman once and for all [1]
> > 
> Good issue. I'm in favour of it. Standards Are Good, especially font
> standards. However, we are probably going to get some gruntling from the
> (active) German and other non-ASCII-language-based Squeak groups...

Latin-1 isn't ASCII (even though the first 128 characters correspond 1-1
with ASCII).  The accented characters and ligatures encoded as 128-255 in
Latin-1 support at least:  French, Spanish, Catalan, Basque, Portuguese,
Italian, Albanian, Rhaeto-Romanic, Dutch, German, Danish, Swedish,
Norwegian, Finnish, Faroese, Icelandic, Irish, Scottish, English,
Afrikaans and Swahili (and hence the entire American continent, Australia,
much of Africa and almost all of continental Europe [plus that miserable
little island off the coast of France, which can't make up its mind about
whether it wants to be part of Europe or part of the USA] ;).

> I'm more than happy to take this on me; although I'm not an expert here

If we adopt Latin-1 fonts within the image then the only remaining
question is whether the "internal" interpretation of glyphs above 127
corresponds to MacRoman (as it does now) or whether it too moves to
Latin-1.  In the first case the fonts will have to be rearranged (to bring
them into line with a total anachronism: the MacOS legacy of Squeak).  In
the second case there's nothing to do because everything "cancels out" and
"just works" -- except in the Mac Classic VM, which would suddenly have to
convert between the internal representation and the external encoding
(which is still MacRoman, since it's running in a MacOS environment).

Maybe some diagrams are in order...

The situation today is something like this:

  platform	internal	external	conversion
  --------	--------	--------	----------
  Mac Classic	MacRoman	MacRoman	none
  MacOSX	MacRoman	Unicode		trivial
  Unix		MacRoman	ISO-8859-1	table lookup
  Win32		MacRoman	no idea		idem

If we moved to unmodified Latin-1 fonts (and Latin-1 interpretation of the
encodings > 127) within the image, we'd have:

  platform	internal	external	conversion
  --------	--------	--------	----------
  Mac Classic	ISO-8859-1	MacRoman	table lookup
  MacOSX	ISO-8859-1	Unicode		trivial
  Unix		ISO-8859-1	ISO-8859-1	none (<grin> ;)
  Win32		ISO-8859-1	no idea		idem

so JMM (and maybe Andreas, I've no idea) and myself would have some VM
tweaking to perform.  (In effect I'd cut some code out of the Unix VM and
email it to JMM for inclusion in the Classic VM ;).

The reasons I don't like the third possibility (Latin-1-based fonts with
MacRoman-like internal encoding) are: (1) the glyphs would have to be
rearranged, and (2) the conversion (in both directions) would be "lossy"
on _all_ platforms since some character encodings >= 128 would not have a
corresponding glyph, and some glyphs would not have a corresponding
character encoding.

(I hope the above is more-or-less correct: I'm not a font expert either.)



More information about the Squeak-dev mailing list