Re: [Emacs-diffs] master c48e539 2/2: Use the expression angle units while simplifying it

[Tino Calancha <tino.calancha@gmail.com>]

On Thu, 25 May 2017, Ken Brown wrote:

> On 5/18/2017 1:47 AM, Tino Calancha wrote:
>> --- a/test/lisp/calc/calc-tests.el
>> +++ b/test/lisp/calc/calc-tests.el
>> @@ -86,6 +86,50 @@ An existing calc stack is reused, otherwise a new one is 
>> created."
>>   					       (math-read-expr "1m") "cm")
>>   			    '(* -100 (var cm var-cm)))))
>>   +(ert-deftest test-calc-23889 ()
>> +  "Test forhttp://debbugs.gnu.org/23889  and 25652."
>> +  (dolist (mode '(deg rad))
>> +    (let ((calc-angle-mode mode))
>> +      ;; If user inputs angle units, then should ignore `calc-angle-mode'.
>> +      (should (string= "5253"
>> +                       (substring
>> +                        (number-to-string
>> +                         (nth 1
>> +                              (math-simplify-units
>> +                               '(calcFunc-cos (* 45 (var rad var-rad))))))
>> +                        0 4)))
>
> This can fail on some platforms, depending on the size of a Lisp integer, 
> because
>
> (nth 1
>     (math-simplify-units
>      '(calcFunc-cos (* 45 (var rad var-rad)))))
>
> doesn't necessarily evaluate to an integer.  For example, on 32-bit Cygwin it 
> evaluates to (bigpos 8818 2198 5253).  This causes the test to fail as 
> follows:
>
> $ make -C test calc-tests SELECTOR='\"test-calc-23889$$\"'
> [...]
> Test test-calc-23889 condition:
>    (wrong-type-argument numberp
>                         (bigpos 8818 2198 5253))
>   FAILED  1/1  test-calc-23889
Thank you.
I pushed a fix which skip the test if the Lisp integer isn't big enough.
It must be OK now.