14:41 <Doob> Hello all. Is the number of bits in a Forth-number dependent on the processor or the standard?

14:41 <Doob> Just i read old books about Forth, where they work with 16 bits everywhere, and then they expand to 32 and 64 by the Forth itself.

15:05 <sts-q> Doob: Well, what i think is, it depends on the implementer. If you implement a Forth, you start with what the machine provides and build what you need. Then extend further.

15:09 <sts-q> Doob: In the good old time ( ;) ) this might be a 8-bit machine, implementing 16-bit Forth in assembly. Then if needed 32-bit Forth in Forth.

15:11 <Doob> sts-q: Ok. This means that if I make a fort in a high-level language with 64-bit numbers, then I do not need to lower the precision and expand the set of words to 16 bits.

15:14 <Doob> Can I use a data stack for floating point numbers? Or is it necessary to work with them in a special float-stack?

15:17 <sts-q> Doob: Floats are data and therefore go on the data-stack, that's fine.

15:21 <sts-q> Doob: I think there are Forth standards: ANSI-something and others. I think, it is a good idea, to start with these, if only for the sake of readability.

15:23 <Doob> I didn't find anything about the bit length, only about memory alignment.

15:26 <sts-q> Doob: In many languages '+' is '+', no matter if int, int32, float, whatever. Forth needs an operator for each type, there are a lot of them.

15:27 <sts-q> http://forth.sourceforge.net/std/fst83/

15:32 <sts-q> http://www.complang.tuwien.ac.at/projects/forth.html

15:32 <Doob> Yeah, but I can hide it from the user. so that the calculation results are correct. I just need to know to what bit size it is critical.

15:32 <Doob> And the existing implementations are confusing. Everyone does it their own way.

15:33 <sts-q> :)) Yes

15:34 <sts-q> bit-size, byte-size or word-size?

15:35 <sts-q> In 'old' books they should have 16-bit-words.

15:36 <sts-q> HTH, need to leave now. Regards.

15:37 <sts-q> (bbiab)