It helps a bit with staging times.

Was previously referred to by name+find, which was wasteful.

This commit adds ?DUP IF guards to MOVE and MOVE- that make them exit
when u is 0. Without these guards a DO loop was executed 65535 times
instead of 0.

In particular, this fixes a crash when "105 LOAD I " is executed
immediately after boot. Block editor word "I" passes IBUF length (0)
to MOVE- and MOVE in this case, causing a crash.

We now have a stable ABI that is pretty cleared up...

The only calls still using the stable ABI indirection are those
made before the next routine is created.

Instead of having wordref point to core word routines, I made them
into word 4 word types. It liberates space into the stable ABI and
should make porting to other arches easier.

I'm also thinking of combining word type with the namelen field
for precious bytes saving, but not now...

It was useful when we still had the relinker, but now it has no use.
I was waiting a bit to see if the distinction would be useful again,
but it seems like it won't.

Only 2>R in actually needed in the RSP family (for LOOP). Reducing
the scope of the stable ABI makes porting to other arches easier.

I had more ambitious goals, but I backed out of it: not worth it.
This small optimization is an easy picking though.

Inline it in next and make chkPS, call a newly reserved label for
it directly, removing a layer of indirection. This frees a spot in
the stable ABI.

Otherwise, when editing a file with VE, we would constantly write
to disk, which is a bit inefficient.

It's a quite useful word and it doesn't make sense to reload it from
disk after each usage.

There wasn't much of a reason to leave a loop early on 0x0d or null.
The idea was that once you have a CR or null, the rest is garbage
and you shouldn't see it.

However, it brought a problem: In VE, you couldn't insert characters
past that limit. It would be written, but never displayed. So let's
get rid of this logic and simply always display a 64x16 grid.

not needed

The address returned in the error condition would be off by one.

I'm not sure why I chose null-terminated initially. Probably because
the z80asm version had null-terminated strings.

Length-prefixes strings are the traditional form of strings in Forth
and it's a bit easier to work with them with traditional forth words
when they're under this form.

I'm planning a string reform and it's standing in the way.

I'm planning on going back to 8-bit branching. 16-bit br cells incur
a non-negligible penalty and, while at first 64 words (128 bytes
forward or backward) seemed a bit limiting, I now don't see why one
would ever construct such a big branch. It would be un-forthy.

Also, I looked at using BC instead of IY to hold IP and the transition
would be a lot easier with 8-bit branching.

In this commit, all I do is add overflow checks in IF. The mechanic
below doesn't change. I'll give myself some time to think it over so
that I avoid yet another back and forth.

Prebiously, when encountering an error during a : ; definition from
input buffer, because the input buffer wasn't flushed, we would continue
interpreting and quit the whole program when encountering ;.

VI, here we come!

Use EXX instead of the stack for HL protection and remove all
spurious uses of chkPS,

I wanted to inline chkPS in next because of its "tight loop" status,
but for reasons I don't understand, doing so breaks Collapse OS.
Later...

It's small enough to be worth it.

It was ill-advised.

I hadn't noticed that this word was almost ANS compliant.

Instead of letting each configuration taking care of RDLN$ and
"CollapseOS" prompt, move this to BOOT to simplify xcomp units.

Initialization source code is now only for driver initialization.

Previously, calling quit would break EOT behavior and not properly
quit Collapse OS.

During "make updatebootstrap", we use less than 0x20 bytes on the
PSP side and less than 0x40 bytes on the RSP one. 0x100 bytes ought
to be enough for anybody.

Use straight VARIABLE instead of Z80MEM+. Initially, I used this
system to allow z80a to be embedded in a system binary, but now
I don't think it's worth it. Compiled, z80a is 2.5k. Sure, it's a
sizeable amount of RAM, but I think that even with it in RAM, I'll
manage a bootstrap within my most constrained machine, the SMS with
8K.

Finally got rid of the XPACKed core and managed to cross-compile
all core words, which greatly simplifies the bootstrapping process.