Just for fun I wanted to see how much memory could be saved with a headerless Forth and I came up with this definition.


0 VARIABLE BSCNTR

: HDRLESS LATEST BEGIN DUP C@ 1F AND 3 + BSCNTR +! PFA LFA @ DUP 0= UNTIL DROP BSCNTR @ . ;

The +3 is for: one byte for the Name Field byte and two for the Link field address, which would not be needed as well.

Currently I could save about 2700 bytes out of 16.4 k used so far. Which is about 16.5%. Doesn't seem like a lot.
The 16.4 k is just strictly memory used by word definitions and does not included memory used by errors, buffers or the User Area.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

I can see a few factors affecting the percentage.
Working on an 8-bit, 16-bit, 32-bit or 64-bit system would have various percentages saved. An 8-bit system with 32-bit support creates a lot of overhead in the main part of the Forth system, which is probably why my percentage is so low. A 16-bit system would probably show a higher percentage of space saved due to a much smaller footprint for the Forth system due to better 16-bit mnemonics.

I suppose in all actuality, one should really only take in account new words added to the Forth system since the words are more likely to be definitions rather than primitives. Which would make the percentage-saved closer among all systems.