On Thu, Feb 16, 2017 at 2:31 PM, Denis Kudriashov <[email protected]>
wrote:

>
> 2017-02-16 13:49 GMT+01:00 Guillermo Polito <[email protected]>:
>
>> I vote for not introducing during code-freeze.
>>
>
> I understand it but these behaviour was introduced in Pharo 6. And It is
> already used quite long time. So it could be considered as small
> improvements to already integrated feature.
>
>
>>
>> We can discuss it for Pharo 7, I'm not against actually. But I have some
>> concerns:
>>
>> - How does it work when we are debugging?
>> I mean, imagine I'm running a test, a debugger is open, and I start
>> working on it. How do these timeout currently work in this scenario? I
>> imagine that the test runner should detect you entered into a "I'm
>> debugging" mode and should not kill your test, but this
>>
>
> It solved very easily. "Watchdog process" just checks that test process
> isSuspended or not. If it is not suspended timeout is signalled. Test
> process is suspended only when debugged.
>

Ok. Is something like this in a class comment? If not, we could add it,
because it will get lost in the mails.


>
>
>> is not clear to me how this works right now.
>>
>> - Also, how can we better document these things? Because I was bitten a
>> couple of times by this timeouts and the "I kill your process but I do not
>> tell you", and it was not nice because it was completely silent or obscure.
>>
>
> Things which you show me I improved. Last days I fix Christophe issues
> which was really important. Anyway "fork management in tests" is not
> related to time limit of tests. And of course it is not supposed to make
> troubles but instead it should make life easy :)
>

I know, I'm not pleading for improvements :). Just talking about
documenttion here.

Reply via email to