Sun Aug 28 17:50:38 MST 2005

trials showed month and weekday had to be lowercase
removed tolower() when passed from  character to block read
stupid...

went back to by character read

[still left in old code and code test fragments, ifdef'd or commented out]

[will have to recompile and reinstall note]

some tests:
chgdate 1/2 +8 m
chgdate JAN 2 +8 m
chgdate -2 1/1/70
chgdate -2 1/1/1970
chgdate -2 1/1/70 -100 y
chgdate -2 1/1/1970 +100 y
chgdate -3 6 3 mon
chgdate -3 6 3 Mon
chgdate -3 6 3 mON
chgdate -3 6 3 mONday
chgdate -3 june 1, 1989 3 mON
chgdate -3 JUNE 1, 1989 3 mO
chgdate now 1 y 1 m 1 w 1 d - now
chgdate now 1 y 1 m 1 w 1 d



Fri Feb  6 23:11:11 MST 2004

found chgdate did not parse input correctly, e.g.:
	chgdate 1/2
		today
		and some others

fixed by revising input == redefine of YY_INPUT
	used simpler = block read inline code

reinstalled executable


also recompiled and reinstalled "note"



Tue Jan 20 08:55:51 MST 2004


recompiled on mac, osx 10.3

problem: gnu yacc and lex differ from att

main change affecting getdate =  no redef of input() in lex
now, use redefine of YY_INPUT - as noted in various refs
done with inline code, as in lex output std define of YY_INPUT
	works with char-by-char read/transfer of input string
	fails if add terminating '\0' at end of transfer to lex buf
does not work with redefine to function, then block read of string
	did not pursue source of error, except
		failure is block read leads to eof seen by lex
		terminating '\0'?
		string is read properly from input buffer sp
works if do not redefine and use yyin to input string

code used to test left in commented out for later play