Re: Test elem1 in the test suite

As author of the test, let me say: The test is meant to check that [] correctly fails to match any input.
The grammar is correct, but it is unable to match any input, since [] never matches. So whatever input it receives, it will fail to find a parse.

The test should succeed if it fails to match its input.

Steven

On Saturday 05 February 2022 15:24:49 (+01:00), C. M. Sperberg-McQueen wrote:

>
> Norm Tovey-Walsh writes:
>
> > Test elem1 is:
>
> > a: "a", b, c.
> > b: "b", c, d.
> > c: "c", []. {it should block here, since nothing matches}
> > d: "d".
>
> > And it’s a syntax test that is suppose to fail:
>
> > <test-case name="elem1">
> > <test-string-ref href="elem1.ixml"/>
> > <result><assert-not-a-sentence/></result>
> > </test-case>
>
> > I interpret the comment to mean that this test should fail because “[]”
> > isn’t valid, but that’s not right:
>
> > -set: -"[", s, member*(-[";|"], s), -"]", s.
>
> > What am I failing to understand about this test?
>
> Good catch.
>
> One possibility is that the test case was formulated before the grammar
> of character sets was changed to allow them to be empty; against the
> then-current ixml specification grammar, it would not be grammatical.
>
> But the phrasing of the comment would be distinctly odd in that case:
> given the old definition of character sets, the natural description of
> [] is that it's an error, not that it matches nothing.
>
> The second possibility that occurs to me is that the test case ended up
> in the wrong directory. It looks to me (by eye) like a perfectly fine
> grammar, so it does not belong in a directory of non-conforming
> grammars. (It would be less of an outlier in Steven's tests/
> directory, I think.)
>
> Since I'm always reluctant to let a good test go to waste, my
> inclination would be to move it to the tests-SP-MSM directory, give it
> some inputs, and specify that none of the inputs are sentences.
>
> My two cents.
>
> Michael
>
>

Received on Tuesday, 8 February 2022 14:56:07 UTC