Bert,

> Actually actually, we patched the StrikeFonts in the image to have an  
> underscore glyph in the place that formerly had an arrow. When we did  
> this, the original arrow glyph was moved to character code 16r8F.

Great - this is the first I heard about this. Which images have it?

When I mentioned "patched fonts" I was talking about way back when the
original Apple fonts were replaced with other with more friendly
licenses and which then had their _ and ^ glyphs changed by hand.
Interesting, the first looks like an underscore here while the second is
an up arrow rather than a caret. Inspecting this PlugglableShoutMorph I
see that this TextStyle is Accuny17. I looked at the SystemBrowser and
it is using Accuny12 and also shows the assignment as underscore. How
odd - I must have been thinking of some other image, then. I use quite a
few different ones.
   
> This code is unassigned in iso8859-1 so pretty safe for private use.

It is the control character "single-shift 3", but it is extremely
unlikely anyone will ever miss it.
  
> If StrikeFonts supported the whole unicode range we could have used  
> 16r2190, the proper unicode codepoint.

That is what I was suggesting (or 16r21D0 for a Smalltalk-74 look), but
it would be very expensive.

> Which character shows the arrow glyph in the currently used font is  
> the thing I think Shout needs to be taught.

All the other tools would have to accept that character as assignment
too, because Shout actually replaces what is typed and doesn't just
change what is displayed.

> I don't think patching the underscore character is the way to go  
> forward. If it appears in code, it should be displayed as an underscore.

As I explained above, I was talking about what happened a long time ago
and not suggesting this as something that should be done now. I just
tried my suggestion in this 3.9 image and it didn't work - every one of
the (rather limited) selection of fonts available shows underscores in
the code browser. In the 3.8 image I use for IRCe it seems that the font
is also Acunny12 but it shows a left arrow instead. The selection of
fonts is exactly the same but it seems that the fonts themselves are
not. Sorry that I didn't test this earlier and so created some
confusion.

About the ":=" as assignment, this was introduced in the first Digitalk
product (called Methods) both to help convert Pascal programmers (their
target audience, if the examples in their manual are any indication) and
due to the hardware limitations of the text mode in CGA boards of the
original PCs. I don't think these reasons matter much today, but being
friendly with other Smalltalks and the ANSI standard is important to me.

Karl,
> 1963 !
> 1967 !!!!
> i can't belive we still are stuck in this mess, then again that is
> probably why we are stuck. Standards, love them or leave them!

The problem was that people couldn't wait for the standard to be
finished to start using it, sort of like the recent 802.11n drama. I did
this twice myself: used SCSI in a 1983 parallel unix machine design and
put the ISO 8859 characters into my 1986 Smalltalk computer. At least
DEC and Xerox (probably many others, but these are the ones I am
familiar with) got stuck with not-fully-ASCII characters in their
equipment, though this machine did more than any other to keep the left
and up arrows alive: http://en.wikipedia.org/wiki/ASR-33_Teletype

-- Jecel