bug#72366: srfi-64: on-bad-end-name-function has swapped arguments

bug#72366: srfi-64: on-bad-end-name-function has swapped arguments

[Tomas Volf]

Hello,

I think I found a bug in (srfi srfi-64) module shipped with GNU Guile.

The specification describes the on-bad-end-name-function like this:

    (on-bad-end-name-function runner begin-name end-name)

Thus the following should print `x':

    (use-modules (srfi srfi-64))
    (let ((r (test-runner-null)))
      (test-runner-current r)
      (test-runner-on-bad-end-name! r
        (λ (runner begin-name end-name)
          (pk begin-name)))
      (test-begin "x")
      (test-assert #t)
      (test-end   "y"))

But it does not:

    ;;; ("y")

Have a nice day
Tomas Volf

bug#72366: srfi-64: on-bad-end-name-function has swapped arguments

[Taylan Kammer]

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 describes the on-bad-end-name-function like this:
>
>     (on-bad-end-name-function runner begin-name end-name)
>
> Thus the following should print `x':
>
>     (use-modules (srfi srfi-64))
>     (let ((r (test-runner-null)))
>       (test-runner-current r)
>       (test-runner-on-bad-end-name! r
>         (λ (runner begin-name end-name)
>           (pk begin-name)))
>       (test-begin "x")
>       (test-assert #t)
>       (test-end   "y"))
>
> But it does not:
>
>     ;;; ("y")
>
> Have a nice day
> Tomas Volf
>
>
This is clearly a bug IMO and I've now fixed it in my implementation.

I've also decided to remove the check on whether the begin-name is true, because as per spec it's not allowed to be false anyway, and IMO it should call the bad-end-name callback if the begin-name was explicitly passed as #f but the end-name is not also #f.

- Taylan

bug#72366: srfi-64: on-bad-end-name-function has swapped arguments

[Taylan Kammer]

[-- Attachment #1: Type: text/plain, Size: 993 bytes --]

On 30.09.2024 16:04, Taylan Kammer wrote:
> I've also decided to remove the check on whether the begin-name is true, because as per spec it's not allowed to be false anyway, and IMO it should call the bad-end-name callback if the begin-name was explicitly passed as #f but the end-name is not also #f.
>
A little update on this:

Removing that check entirely was stupid, because the value defaults to #f when not provided. So, an explicit #f is allowed by `test-end` as an implementation detail, and is equivalent to not providing a name. Thus, starting a test suite with a name "foo" and ending it with explicit name #f is valid.

However, I've changed the code of `test-end` to check whether the `end-name` was provided rather than whether the `begin-name` was provided. Thus, starting a test suite with an explicit name #f and ending it with a non-#f name WILL raise an error. If `test-end` is provided a name and it is non-#f, then it MUST match the name passed to `test-begin`.

- Taylan

[-- Attachment #2: Type: text/html, Size: 1493 bytes --]