On 30.07.2024 21:51, Tomas Volf wrote:
> Hello,
>
> I think I found a bug in (srfi srfi-64) module shipped with GNU Guile.
>
> The specification says the following regarding the test-approximate:
>
>> This is equivalent to (except that each argument is only evaluated once):
>>
>>
>> (test-assert [test-name]
>> (and (>= test-expr (- expected error))
>> (<= test-expr (+ expected error))))
> However the test-name is evaluated multiple times:
>
> (use-modules (srfi srfi-64))
> (test-begin "x")
> (test-approximate (pk "t") 1 1 0)
>
> Leading to:
>
> ;;; ("t")
>
> ;;; ("t")
>
> Have a nice day
> Tomas Volf
>
>
Looks like a bug in the reference implementation that Guile uses. It
specifically evaluates the macro input `tname` at the beginning, binding it to
the variable `name`, but then never uses `name` and just uses `tname` again,
leading to it being evaluated twice.
If someone wants to fix it upstream, it should be easy: Just change `tname` to
`name` on line 788.
My implementation already seems to have this fixed, because the affected
section of the code was entirely rewritten. (In case someone is missing the
context: I've explained in a previous email that I don't want to work on the
reference implementation / the one shipped with Guile, because I find it too
unreadable and difficult to work with, although in this particular case it
looks like a trivial bug.)
- Taylan
