suzanne.soy/racket
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

racket/fasl: fixed representation for +nan.0

Browse Source 
To make fasl writing as determinsitic and portable as possible, write
+nan.0 and +nan.f always with a specific bit pattern.

This choice risks losing information that is potentially useful, but
given the way that Racket treats all NaN encodings as equivalent, that
rick seems low.
This commit is contained in:
Matthew Flatt 2019-03-12 16:06:55 -06:00
parent 6a35d64e95
commit e79f07b6cc
1 changed files with 10 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
racket/collects/racket/fasl.rkt
View File

@ -195,10 +195,18 @@
(write-fasl-integer v o)])]
[(flonum? v)
(write-byte fasl-flonum-type o)
(write-bytes (real->floating-point-bytes v 8 #f) o)]
(write-bytes (if (eqv? v +nan.0)
;; use a canonical NaN (0 mantissa)
#"\0\0\0\0\0\0\370\177"
(real->floating-point-bytes v 8 #f))
o)]
[(single-flonum? v)
(write-byte fasl-single-flonum-type o)
(write-bytes (real->floating-point-bytes v 4 #f) o)]
(write-bytes (if (eqv? v +nan.f)
;; use a canonical NaN (0 mantissa)
#"\0\0\300\177"
(real->floating-point-bytes v 4 #f))
o)]
[(extflonum? v)
(write-byte fasl-extflonum-type o)
(define bstr (string->bytes/utf-8 (format "~a" v)))