Please add some sugar for reporting details from a fixture

[James Westby]

Hi,

When using fixtures they sometimes have some data that I want to see, usually
on failures, but perhaps any time. It would be great if they could easily supply
that data to testtools for reporting through it's details API.

The example I have right now is the log file from a server running in a different
process.

Thanks,

James

[Robert Collins (lifeless)]

I suggest a simple details() call which should return an iterable or dict (perhaps dict ?) of details objects.

[James Westby (james-w)]

On Sat, 18 Sep 2010 03:02:55 -0000, Robert Collins <email address hidden> wrote:
> I suggest a simple details() call which should return an iterable or
> dict (perhaps dict ?) of details objects.

I would think get_details like mismatches, which returns a dict?

Would there be a use case for controlling whether the details are
attached only on failure, or always?

Thanks,

James

[Robert Collins (lifeless)]

On Sun, Sep 19, 2010 at 2:30 AM, James Westby <email address hidden> wrote:
> On Sat, 18 Sep 2010 03:02:55 -0000, Robert Collins <email address hidden> wrote:
>> I suggest a simple details() call which should return an iterable or
>> dict (perhaps dict ?) of details objects.
>
> I would think get_details like mismatches, which returns a dict?
>
> Would there be a use case for controlling whether the details are
> attached only on failure, or always?

I think that that would apply to the test not the fixture? or perhaps
for some fixtures you want it different?

[James Westby (james-w)]

On Sat, 18 Sep 2010 20:25:19 -0000, Robert Collins <email address hidden> wrote:
> On Sun, Sep 19, 2010 at 2:30 AM, James Westby <email address hidden> wrote:
> > On Sat, 18 Sep 2010 03:02:55 -0000, Robert Collins <email address hidden> wrote:
> >> I suggest a simple details() call which should return an iterable or
> >> dict (perhaps dict ?) of details objects.
> >
> > I would think get_details like mismatches, which returns a dict?
> >
> > Would there be a use case for controlling whether the details are
> > attached only on failure, or always?
>
> I think that that would apply to the test not the fixture? or perhaps
> for some fixtures you want it different?

Maybe. Generally you only care on failure, but it is something that is
different to the mismatch interface, as they are only queried on
failure.

Thanks,

James

[Robert Collins (lifeless)]

Sure; I think mainly I'm reticent to teach Fixtures about
success/failure. It seems a different concern to me.