Mailing-List: contact users-help@maven.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "Maven Users List" <users@maven.apache.org>
Received-SPF: pass (athena.apache.org: domain of kjkrum@gmail.com designates
 209.85.160.178 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <008101cf2834$fcf8d4d0$f6ea7e70$@berthonneau.com>
References: <9e5201cf2831$52f17c60$f8d47520$@berthonneau.com>
	<CABLGb9yyWxO8om2YE2Y-3v0RwLgB4RmY6cG9UK330DY4UVGHpQ@mail.gmail.com>
	<008101cf2834$fcf8d4d0$f6ea7e70$@berthonneau.com>
Date: Wed, 12 Feb 2014 14:10:31 -0700
Message-ID: 
 <CAFBdXRp0KqmGHnDjjD3=_U94nh4HNo1zWrV5KNVRRxDeVSBjoQ@mail.gmail.com>
Subject: RE: Code coverage with debug logs: 100% branch coverage not
 possible?...
From: Kevin Krumwiede <kjkrum@gmail.com>
To: Maven Users List <users@maven.apache.org>
Content-Type: multipart/alternative; boundary=20cf303a2d8bbe86e804f23c00ff

--20cf303a2d8bbe86e804f23c00ff
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

It does matter which implementation.  The main reason it was recommended to
check the logging level was because string concatenation can be expensive,
and you want to avoid doing it for a message that won't be logged.  But if
you're using a logging API like slf4j that uses parameter replacement
tokens in the message string, if the message isn't logged, the replacement
won't be performed and the call will be cheap.
On Feb 12, 2014 1:57 PM, "Beno=EEt Berthonneau" <benoit@berthonneau.com>
wrote:

> Hi Paul,
>
>
>
> Don't think that I could play with exclusions. Here is an example :
>
>
>
> *A Unit Test :*
>
>
>
> *The tested class with ALL traces activated:*
>
>
>
> *And the same tested class with INFO traces activated:*
>
>
>
>
>
> -----Message d'origine-----
> De : paulus.benedictus@gmail.com [mailto:paulus.benedictus@gmail.com] De
> la part de Paul Benedict
> Envoy=E9 : mercredi 12 f=E9vrier 2014 21:36
> =C0 : Maven Users List
> Objet : Re: Code coverage with debug logs: 100% branch coverage not
> possible?...
>
>
>
> IIRC, there should be an option in Emma/Cobertura that allows you to
> exclude coverage on certain classes. So if you can exclude your log4j
> classes (you don't really want to test your logging, do you?), then you
> should be able to raise your percentage.
>
>
>
>
>
> On Wed, Feb 12, 2014 at 2:30 PM, Beno=EEt Berthonneau
>
> <benoit@berthonneau.com>wrote:
>
>
>
> > Hi all,
>
> >
>
> >
>
> >
>
> > I need your opinion/way to tackle the following problem:
>
> >
>
> > In many projects we use a Logger (doesn't matter which
>
> > implementation). It is often recommend to test if the debug level is
>
> > activated before logging a debug trace like the following:
>
> >
>
> > if (logger.isDebugEnabled()) {
>
> >
>
> >     logger.debug("blah " + i + " in the loop that contains " + max);
>
> >
>
> > }
>
> >
>
> >
>
> >
>
> > Now when you run unit tests on this kind of code you need to make a
> choice:
>
> > run tests with INFO level or run tests with ALL traces activated. I
>
> > choose the second option in order to:
>
> >
>
> > *         Check that debug traces doesn't throw unwanted exception (lik=
e
>
> > NPE)
>
> >
>
> > *         Have a better code coverage in term of covered lines
>
> >
>
> >
>
> >
>
> > But in term of branches coverage we could never have a 100% :(
>
> >
>
> >
>
> >
>
> > To me the only way to cover this is to run the tests suite 2 times:
>
> > one with INFO traces configured, and another one with ALL traces
>
> > activated.
>
> >
>
> > Did you face this issue and how did you solve it ?
>
> >
>
> >
>
> >
>
> > Thanks,
>
> >
>
> > Beno=EEt.
>
> >
>
> >
>
>
>
>
>
> --
>
> Cheers,
>
> Paul
>

--20cf303a2d8bbe86e804f23c00ff--