Integration vs Unit tests

[Maurizio Cucchiara <mc...@apache.org>]

Personally I like very much junit-benchmark  (thanks to Raffaele who
let me know about it), I also understand that accessing to memory
might produce concurrency issues, but I think that integration tests
provide many benefits that unit tests do not (and vice versa).
Moreover, benchmarks are much more expansive in terms of
time/resources consume.
Slow tests give a slow feedback and in the long run might discourage
developers from running tests.
So I'd love to keep them separate and maybe introduce a maven
profile/var to skip benchmarks.

WDYT?

Twitter     :http://www.twitter.com/m_cucchiara
G+          :https://plus.google.com/107903711540963855921
Linkedin    :http://www.linkedin.com/in/mauriziocucchiara

Maurizio Cucchiarae

Re: Integration vs Unit tests

[Maurizio Cucchiara <mc...@apache.org>]

Thank Raffaele for your feedback.
Run integration tests on nightly build is a good idea.
>should fail if performance degrades
This is also a good idea, personally I have not a big experience in
this field, we can consider to use timeout parameter of junit, but I
think that execution time is related to the resource business (Jenkins
instances are usually shared with other projects). Maybe there is a
jenkins plugin, some jenkins guru might give us the right answer :).


FYI I have just filed an issue on this topic
https://issues.apache.org/jira/browse/DIRECTMEMORY-51

Twitter     :http://www.twitter.com/m_cucchiara
G+          :https://plus.google.com/107903711540963855921
Linkedin    :http://www.linkedin.com/in/mauriziocucchiara

Maurizio Cucchiara



On 31 December 2011 00:12, Raffaele P. Guidi <ra...@gmail.com> wrote:
> A good idea. I used to have a completely separate project for benchmarks,
> with an older implementation. I to think these should run nightly instead
> than continuously and should fail if performance degrades - but at the time
> I didn't have a CI server.
>
> +1
> Il giorno 30/dic/2011 22:58, "Maurizio Cucchiara" <mc...@apache.org>
> ha scritto:
>
>> Personally I like very much junit-benchmark  (thanks to Raffaele who
>> let me know about it), I also understand that accessing to memory
>> might produce concurrency issues, but I think that integration tests
>> provide many benefits that unit tests do not (and vice versa).
>> Moreover, benchmarks are much more expansive in terms of
>> time/resources consume.
>> Slow tests give a slow feedback and in the long run might discourage
>> developers from running tests.
>> So I'd love to keep them separate and maybe introduce a maven
>> profile/var to skip benchmarks.
>>
>> WDYT?
>>
>> Twitter     :http://www.twitter.com/m_cucchiara
>> G+          :https://plus.google.com/107903711540963855921
>> Linkedin    :http://www.linkedin.com/in/mauriziocucchiara
>>
>> Maurizio Cucchiarae
>>

Re: Integration vs Unit tests

["Raffaele P. Guidi" <ra...@gmail.com>]

A good idea. I used to have a completely separate project for benchmarks,
with an older implementation. I to think these should run nightly instead
than continuously and should fail if performance degrades - but at the time
I didn't have a CI server.

+1
Il giorno 30/dic/2011 22:58, "Maurizio Cucchiara" <mc...@apache.org>
ha scritto:

> Personally I like very much junit-benchmark  (thanks to Raffaele who
> let me know about it), I also understand that accessing to memory
> might produce concurrency issues, but I think that integration tests
> provide many benefits that unit tests do not (and vice versa).
> Moreover, benchmarks are much more expansive in terms of
> time/resources consume.
> Slow tests give a slow feedback and in the long run might discourage
> developers from running tests.
> So I'd love to keep them separate and maybe introduce a maven
> profile/var to skip benchmarks.
>
> WDYT?
>
> Twitter     :http://www.twitter.com/m_cucchiara
> G+          :https://plus.google.com/107903711540963855921
> Linkedin    :http://www.linkedin.com/in/mauriziocucchiara
>
> Maurizio Cucchiarae
>