[bug#35929] [PATCH] tests: hackage: avoid mock, and extract test data

[Ludovic Courtès]

Hi,

Robert Vollmert <address@hidden> skribis:

>> On 29. May 2019, at 23:16, Ludovic Courtès <address@hidden> wrote:
>> --8<---------------cut here---------------start------------->8---
>> scheme@(guile-user)> (define-syntax-rule (match-pattern obj pattern)
>>                     (match obj
>>                       (pattern #t)
>>                       (_       #f)))
>> scheme@(guile-user)> (match-pattern '(1 2 3) (a b c))
>> $14 = #t
>> scheme@(guile-user)> (match-pattern '(1 2 3) (a b))
>> $15 = #f
>> scheme@(guile-user)> (match-pattern '(7 7) (a a))
>> $16 = #t
>> scheme@(guile-user)> (match-pattern '(1 2) (a a))
>> $17 = #f
>> --8<---------------cut here---------------end--------------->8—
>
> Hmm, I can’t quite get this to work. I’d like to supply a name for
> the pattern, so I can reuse it across test cases, and I’m currently
> at a loss for how to do tha:
>
> scheme@(ice-9 match)> (define pattern '(a b c))
> scheme@(ice-9 match)> (match-pattern '(a 2) pattern)
> $30 = #t

Indeed, ‘match’ expects patterns to be literals.  So you cannot do:

  (define pattern …)

and then use that pattern in ‘match’ clauses.

If you wanted to do that, you would need to define a “procedural”
variant of ‘match’, which is not terribly difficult either, but it’s a
different beast.

> I did manage to get things to work using eval, will send a revised
> patch.

Please don’t use ‘eval’; there’s a saying that “eval is evil”, because
basically anything can happen when calling out to ‘eval’.

So as it turns out, I realize that the factorization you wanted to make
is not all that easy to achieve.  Perhaps adding a procedural pattern
matcher, which would be similar to ‘equal?’ except that it would ignore
the sha256, synopsis, and description, is the right way to go.

HTH!

Ludo’.