[104] 2019-08-05 17:25:38 flaagel wrote:

referenceImplementation - GNU C RESTRICT would make sense in the standard

IMMEDIATE sets the corresponding flag in the latest compiled word header. GNU Forth RESTRICT sets the compile-only flag in a similar manner. I think it should be standardized for the sake of completeness.

[r273] 2019-08-05 07:13:43 AntonErtl replies:

requestClarification - Test example numeric base?

Hayes' core.fr starts with HEX. Now that we have standardized number prefixes, we could use number prefixes (e.g., $20) to eliminate this dependence on BASE.

[r274] 2019-08-05 15:55:56 ruv replies:

requestClarification - Test example numeric base?

I think, regardless other third party files, numbers in the specification should be in decimal by default. The test suit from Annex F restores BASE after loading.

Also, in other places the decimal radix is implied. E.g. the tests for +LOOP.

[r275] 2019-08-06 09:28:24 ruv replies:

referenceImplementation - GNU C RESTRICT would make sense in the standard

restrict in Gforth is synonym for compile-only. Its behavior is described as "Remove the interpretation semantics of a word". Such behavior description cannot take place in the standard.

By the standard, if a word with undefined interpretation semantics is encountered by the text interpreter in interpretation state the occurring behavior is implementation specific. A possible variant is to even perform its compilation semantics. There is no much sense for a standard program to transform the well defined interpretation semantics of some word to something unknown, that will be depend on implementation.