Congratulations on Gforth 0.7.0! I'd like to propose the following revision to the compilation logic in the CVS version (revision 1.11) of ttester.fs:
http://www-personal.umich.edu/~williams/archive/forth/utilities/ttester-20081031.fs The problem I had with the original version was that it wouldn't load for systems with no floating point, as in the default state for pfe. Such a system tries and fails to compile the [ELSE]...[THEN] clause in lines 209--230 (watch out for line wrap): HAS-FLOATING-STACK [IF] ... [ELSE] : EMPTY-FSTACK ; : F{ ; : F-> ; : F} ; : F...}T ; DECIMAL : COMPUTE-CELLS-PER-FP ( -- U ) DEPTH 0E DEPTH 1- >R FDROP R> SWAP - ; HEX COMPUTE-CELLS-PER-FP CONSTANT CELLS-PER-FP : FTESTER ( R -- ) DEPTH CELLS-PER-FP < ACTUAL-DEPTH @ XCURSOR @ START-DEPTH @ + CELLS-PER-FP + < OR IF S" NUMBER OF RESULTS AFTER '->' BELOW ...}T SPECIFICATION: " ERROR EXIT ELSE ACTUAL-RESULTS XCURSOR @ CELLS + F@ FCONF= 0= IF S" INCORRECT FP RESULT: " ERROR THEN THEN CELLS-PER-FP XCURSOR +! ; [THEN] That belies the statement in lines 39 and 40: \ possible; e.g., FLOATING words are used only if the FLOATING wordset \ is present The proposed revision modifies the compilation logic so that floating point words are not used when HAS-FLOATING is false, and so that the mixed stack words starting with ...}T are then simply omitted. The indentation gets a bit nasty. :-( I've used it with pfe and gforth for * tests for a fairly complex application not involving floating point * Krishna myneni's fsl test suite (see the .tar.gz at ftp://ccreweb.org/software/fsl/) For pfe, I've also done rudimentary tests with floating point and an integrated stack. As far as I can tell, that covers the logic revisions. -- David P.S. I wasn't sure what to do about a change log entry for the revision in the file itself, so I did nothing. It appears that, since the first few revisions, changes have not been logged in the file, but rather are covered by CVS. Should there be some statement in the file about how to find the change record? --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
