|
|
Boost Testing : |
From: Thomas Witt (witt_at_[hidden])
Date: 2007-04-01 14:15:50
Hi,
David Abrahams wrote:
> on Fri Mar 30 2007, Martin Wille <mw8329-AT-yahoo.com.au> wrote:
>
>> David Abrahams wrote:
>>> on Thu Mar 29 2007, Martin Wille <mw8329-AT-yahoo.com.au> wrote:
>>>
>>>> IF all developers tell the testers about removed or renamed tests AND IF
>>>> the incremental testers then remove the old results then that doesn't
>>>> happen.
>>>>
>>>> We could discuss adding formal support for that communication to
>>>> Boost.Build. However, I currently see this problem as behaviour-related
>>>> or knowledge-related, not as tool-related.
>>> If the reporting tools aren't removing the tests from the tables, they
>>> should be, and it's a tool issue... but I don't think the answer is to
>>> ban incremental testing.
>>
>> IMHO, fighting against invalid data is better to do early in a tool
>> chain.
I am willing to accept the argument that non-incremental testing is
simply not viable for performance reasons.
That being said unless you guys volunteer to fix the tools I am in
strong disagreement with everything else that was said.
It is true that in theory what we have works IF everybody does the right
thing, but that's beside the point. My point is whether it works in
practice and I do believe that there is ample evidence that it does not.
If every removed test requires several emails going back and forth,
people getting LOUD and days of turnaround time then to me there is
something wrong. As it is the system is too brittle.
>
> bjam already spits out a list of all the tests at the beginning of a
> run; how much earlier can we do it?
The system might be fixable but unless somebody has the resources to do
it it's a moot point.
Thomas
-- Thomas Witt witt_at_[hidden]