[Schematics-development] Re: SRFI 19 ~f bug

[David V. H. <dva...@cs...>]

Noel Welsh wrote:
> Thanks for the bug report.  It would be helpful it you
> could give the expected behaviour as well -- I can't keep
> up-to-date with all Schematics projects.

The description of "~f" is:
"seconds+fractional seconds, using locale's decimal separator (e.g. 5.2)."

There is no specification of the precision of the fractional seconds, so it 
seems the easiest implementation is to do the following, ie use all 9 digits 
available:

(string-append
   (string->number (date-seconds d))
   (locale-separator)
   (string-pad (string->number (date-nanoseconds d)) 9 #\0))

But any of the following would be fine according to the SRFI, as far as I can 
tell (assuming the locale decimal separator is "."):

(define d (time-utc->date (make-time time-utc 5290000 0)))
(date->string d "~f")
=>
"0.5"
"0.53"
"0.529"
"0.5290"
"0.52900"
...

PLT gives:
(date->string d "~f")
=>
"00.01-53-1"

However, there are numerous other problems with the SRFI reference 
implementation which have leaked over into the PLT version.  I'm going to 
rewrite it from scratch in hopes of fixing it once and for all, so don't worry 
about fixing this particular bug for the moment.

> PS: I'm about to commit your SRFI 60 impl to Schematics
> CVS.

Thanks.

David