2 patches for repository http://common-lisp.net/project/cl-json/darcs/cl-json:
Sun Nov 4 15:49:32 CST 2012 rpgoldman@sift.info * fix-fiveam-test-comparisons
FiveAM would like the expected value to be before the computed value, for best printing when tests fail. I reversed the arguments to the JSON-NUMBER test to agree with this convention.
Sun Nov 4 15:52:54 CST 2012 rpgoldman@sift.info * remove-read-from-string-from-parse-number
PARSE-NUMBER used to use READ-FROM-STRING. This was causing problems with some application-building systems. We replace use of the reader (through read-from-string) by PARSE-NUMBER.
On Sun, Nov 4, 2012 at 10:57 PM, rpgoldman@sift.info wrote:
2 patches for repository http://common-lisp.net/project/cl-json/darcs/cl-json:
Sun Nov 4 15:49:32 CST 2012 rpgoldman@sift.info
- fix-fiveam-test-comparisons
Sun Nov 4 15:52:54 CST 2012 rpgoldman@sift.info
- remove-read-from-string-from-parse-number
Applied, thanks ! /Henrik
On 11/5/12 Nov 5 -3:04 PM, Henrik Hjelte wrote:
On Sun, Nov 4, 2012 at 10:57 PM, <rpgoldman@sift.info mailto:rpgoldman@sift.info> wrote:
2 patches for repository http://common-lisp.net/project/cl-json/darcs/cl-json: Sun Nov 4 15:49:32 CST 2012 rpgoldman@sift.info <mailto:rpgoldman@sift.info> * fix-fiveam-test-comparisons Sun Nov 4 15:52:54 CST 2012 rpgoldman@sift.info <mailto:rpgoldman@sift.info> * remove-read-from-string-from-parse-numberApplied, thanks ! /Henrik
Thanks. It would be great if someone could test CL-JSON on ECL, both before and after my modification.
I don't know if my modification causes ECL to fail to pass its tests, or whether (as I suspect), it reveals problems with the
I am embarrassed to reveal that darcs is all busted on my machine, and my attempts to revert CL-JSON simply resulted in my working copy getting wiped, and two bug reports to darcs... So I'm not capable of checking this myself.
One patch I should have applied, but didn't, was a bump to CL-JSON's patch level.... Maybe someone could do that.
Best, r
Ah, crumbs -- pressed "send" too soon.
What I meant to say was that I couldn't tell if I broke the float tests on ECL or, as I suspect, they weren't passing before, because of float comparison errors.
I don't know enough about the differences between ECL's implementation of floating point and that used in other CL implementations to hazard a guess.
And I can't tell because my working copy is destroyed.... :-(
-
Henrik Hjelte -
Robert Goldman -
rpgoldman＠sift.info