Re: [Guile-commits] GNU Guile branch, stable-2.0, updated. v2.0.5-68-gdab48cc

[Ludovic Courtès]

Hi,

Andy Wingo <address@hidden> skribis:

> On Tue 06 Mar 2012 18:13, address@hidden (Ludovic Courtès) writes:
>
>> "Andy Wingo" <address@hidden> skribis:
>>
>>> commit dab48cc567f931b275ad647db1e47135b63c6675
>>> Author: Andy Wingo <address@hidden>
>>> Date:   Fri Mar 2 17:46:28 2012 +0100
>>>
>>>     tweaks to -Wformat's gettext detection
>>>     
>>>     * module/language/tree-il/analyze.scm (proc-ref?): Change to use less
>>>       false-if-exception and more variable-bound?.  If a variable is present
>>>       in the local module but not bound, assume that it is gettext if it has
>>>       the right name.  This is to allow for (define _ gettext).
>>
>> What’s the functional change?
>
> The previous behavior depended on the side effect of the expander
> producing an unbound variable in the current module.  This behavior is
> not present in master.  I changed the test case so as to do something
> that would work in master.

Oh, I understand now.  Thanks for the explanation.

>> Could you reinstate "non-literal format string with forward
>> declaration", which tests something different as the name implies (see
>> d3160473)?
>
> Sure.  First, though, would you mind discussing the desired behavior a
> bit more?

Yep.

The test was added as part of d3160473, which fixed ‘proc-ref?’; before
that commit, this test would fail with an error.

> I think the "forward declaration" test was not quite right, because it
> would not work with a local (define _ gettext).  That's why I changed
> the test from checking that a warning was issued on a local (define _
> ...) to change that a warning was not issued on that same case.

Hmm, I see.  Before, in stable-2.0, it would see ‘_’ as an unbound
variable, and ‘gettext?’ would return #f, instead of #f, right?

Thanks,
Ludo’.