Quantcast

Coverage seems to have dropped since 3.0.1

classic Classic list List threaded Threaded
9 messages Options
dbb
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Coverage seems to have dropped since 3.0.1

dbb
I have had a few teams approach me lately about coverage numbers on their projects being lower.
Without any code or tests ran being changed, just re-builds/re-analysis, the coverage percent has lowered.
This was noticed from going from 3.0.1 to 3.1.1 and still in the 3.2 release.
I have not seen any e-mail on the list about this specifically.
Has anyone else noticed this?

Doug
--

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email


Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Coverage seems to have dropped since 3.0.1

Patroklos Papapetrou
Hi Doug
The only reason I can think that your code coverage has been lowered when analysing exactly the same code base is the change of coverage tool. Since 3.2 the default code coverage tool of Sonar is Jacoco so this might be the reason but I can't understand why this happened in 3.1.1. Are you sure you haven't changed anything?
Regards
Patroklos

2012/8/14 Doug Beattie <[hidden email]>
I have had a few teams approach me lately about coverage numbers on their projects being lower.
Without any code or tests ran being changed, just re-builds/re-analysis, the coverage percent has lowered.
This was noticed from going from 3.0.1 to 3.1.1 and still in the 3.2 release.
I have not seen any e-mail on the list about this specifically.
Has anyone else noticed this?

Doug
--

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email



dbb
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Coverage seems to have dropped since 3.0.1

dbb
Patroklos,
Thanks for the response.
The drop was noticed when I switched from 3.0.1 to 3.1.1. (I skipped going to 3.1).
From what the teams have said they have not changed any code or tests.
Metrics for the LOC/branches, etc. and tests appear to be the same.
I also wondered about JaCoCo but that wouldn't have necessarily have come into play until 3.2 although they may have switch to using it with the 3.1.1 release.
I'll check into this more.
BTW, if the tests are the same why would JaCoCo give different coverage results than Cobetura?
If they are different which one was more correct? (Just wondering.)

Doug


On Wed, Aug 15, 2012 at 10:10:29AM +0300, Papapetrou P.Patroklos wrote:

> Hi Doug
> The only reason I can think that your code coverage has been lowered when
> analysing exactly the same code base is the change of coverage tool. Since
> 3.2 the default code coverage tool of Sonar is Jacoco so this might be the
> reason but I can't understand why this happened in 3.1.1. Are you sure you
> haven't changed anything?
> Regards
> Patroklos
>
> 2012/8/14 Doug Beattie <[hidden email]>
>
> > I have had a few teams approach me lately about coverage numbers on their
> > projects being lower.
> > Without any code or tests ran being changed, just re-builds/re-analysis,
> > the coverage percent has lowered.
> > This was noticed from going from 3.0.1 to 3.1.1 and still in the 3.2
> > release.
> > I have not seen any e-mail on the list about this specifically.
> > Has anyone else noticed this?
> >
> > Doug
> > --
> >
> > ---------------------------------------------------------------------
> > To unsubscribe from this list, please visit:
> >
> >     http://xircles.codehaus.org/manage_email
> >
> >
> >

--

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email


dbb
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Coverage seems to have dropped since 3.0.1

dbb
It appears the coverage percent drop may have started with 3.2 and not 3.1.
Those reporting to me have all been giving dates that correspond with when I brought up 3.2
I've attached a screen capture which shows a project with no code change and the same tests ran,
but it shows a marked drop in coverage that corresponds with when I updated to 3.2
I did the update late night on 11 August, 2012, This would have been early morning UTC on the 12th.
Again, I asked if anyone else has seen this behavior?
Also, what can I do to better analyze why the drop?

Thanks for any suggestions.

Doug

On Wed, Aug 15, 2012 at 08:23:11AM -0600, Doug Beattie wrote:

> Patroklos,
> Thanks for the response.
> The drop was noticed when I switched from 3.0.1 to 3.1.1. (I skipped going to 3.1).
> From what the teams have said they have not changed any code or tests.
> Metrics for the LOC/branches, etc. and tests appear to be the same.
> I also wondered about JaCoCo but that wouldn't have necessarily have come into play until 3.2 although they may have switch to using it with the 3.1.1 release.
> I'll check into this more.
> BTW, if the tests are the same why would JaCoCo give different coverage results than Cobetura?
> If they are different which one was more correct? (Just wondering.)
>
> Doug
>
>
> On Wed, Aug 15, 2012 at 10:10:29AM +0300, Papapetrou P.Patroklos wrote:
> > Hi Doug
> > The only reason I can think that your code coverage has been lowered when
> > analysing exactly the same code base is the change of coverage tool. Since
> > 3.2 the default code coverage tool of Sonar is Jacoco so this might be the
> > reason but I can't understand why this happened in 3.1.1. Are you sure you
> > haven't changed anything?
> > Regards
> > Patroklos
> >
> > 2012/8/14 Doug Beattie <[hidden email]>
> >
> > > I have had a few teams approach me lately about coverage numbers on their
> > > projects being lower.
> > > Without any code or tests ran being changed, just re-builds/re-analysis,
> > > the coverage percent has lowered.
> > > This was noticed from going from 3.0.1 to 3.1.1 and still in the 3.2
> > > release.
> > > I have not seen any e-mail on the list about this specifically.
> > > Has anyone else noticed this?
> > >
> > > Doug
> > > --
> > >
> > > ---------------------------------------------------------------------
> > > To unsubscribe from this list, please visit:
> > >
> > >     http://xircles.codehaus.org/manage_email
> > >
> > >
> > >
>
> --
>
> ---------------------------------------------------------------------
> To unsubscribe from this list, please visit:
>
>     http://xircles.codehaus.org/manage_email
>
>

--

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email


dbb
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Coverage seems to have dropped since 3.0.1

dbb
Forgot to attached the screen shot. Here it is.

On Fri, Aug 17, 2012 at 08:29:54AM -0600, Doug Beattie wrote:

> It appears the coverage percent drop may have started with 3.2 and not 3.1.
> Those reporting to me have all been giving dates that correspond with when I brought up 3.2
> I've attached a screen capture which shows a project with no code change and the same tests ran,
> but it shows a marked drop in coverage that corresponds with when I updated to 3.2
> I did the update late night on 11 August, 2012, This would have been early morning UTC on the 12th.
> Again, I asked if anyone else has seen this behavior?
> Also, what can I do to better analyze why the drop?
>
> Thanks for any suggestions.
>
> Doug
>
> On Wed, Aug 15, 2012 at 08:23:11AM -0600, Doug Beattie wrote:
> > Patroklos,
> > Thanks for the response.
> > The drop was noticed when I switched from 3.0.1 to 3.1.1. (I skipped going to 3.1).
> > From what the teams have said they have not changed any code or tests.
> > Metrics for the LOC/branches, etc. and tests appear to be the same.
> > I also wondered about JaCoCo but that wouldn't have necessarily have come into play until 3.2 although they may have switch to using it with the 3.1.1 release.
> > I'll check into this more.
> > BTW, if the tests are the same why would JaCoCo give different coverage results than Cobetura?
> > If they are different which one was more correct? (Just wondering.)
> >
> > Doug
> >
> >
> > On Wed, Aug 15, 2012 at 10:10:29AM +0300, Papapetrou P.Patroklos wrote:
> > > Hi Doug
> > > The only reason I can think that your code coverage has been lowered when
> > > analysing exactly the same code base is the change of coverage tool. Since
> > > 3.2 the default code coverage tool of Sonar is Jacoco so this might be the
> > > reason but I can't understand why this happened in 3.1.1. Are you sure you
> > > haven't changed anything?
> > > Regards
> > > Patroklos
> > >
> > > 2012/8/14 Doug Beattie <[hidden email]>
> > >
> > > > I have had a few teams approach me lately about coverage numbers on their
> > > > projects being lower.
> > > > Without any code or tests ran being changed, just re-builds/re-analysis,
> > > > the coverage percent has lowered.
> > > > This was noticed from going from 3.0.1 to 3.1.1 and still in the 3.2
> > > > release.
> > > > I have not seen any e-mail on the list about this specifically.
> > > > Has anyone else noticed this?
> > > >
> > > > Doug
> > > > --
> > > >
> > > > ---------------------------------------------------------------------
> > > > To unsubscribe from this list, please visit:
> > > >
> > > >     http://xircles.codehaus.org/manage_email
> > > >
> > > >
> > > >
> >
> > --
> >
> > ---------------------------------------------------------------------
> > To unsubscribe from this list, please visit:
> >
> >     http://xircles.codehaus.org/manage_email
> >
> >
>
> --
>
> ---------------------------------------------------------------------
> To unsubscribe from this list, please visit:
>
>     http://xircles.codehaus.org/manage_email
>
>
--


---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

CoverageDropWith3-2.png (161K) Download Attachment
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Coverage seems to have dropped since 3.0.1

David Racodon-2
Hi Doug,

There could be two different explanations:
  1. As Patroklos said, since Sonar 3.2, the default code coverage tool is JaCoCo. So, if no coverage tool was explicitly defined through the 'sonar.core.codeCoveragePlugin' property, a switch between Cobertura and JaCoCo has happened. It would explain the differences.
  2. If the code coverage tool was previously JaCoCo, the code coverage can be slighty different as well because JaCoCo was upgraded from version 0.5.6 to 0.5.8. See http://docs.codehaus.org/display/SONAR/Plugin+Versions+Matrix.
Regards,

David RACODON | SonarSource
Senior Consultant



On 17 August 2012 16:32, Doug Beattie <[hidden email]> wrote:
Forgot to attached the screen shot. Here it is.

On Fri, Aug 17, 2012 at 08:29:54AM -0600, Doug Beattie wrote:
> It appears the coverage percent drop may have started with 3.2 and not 3.1.
> Those reporting to me have all been giving dates that correspond with when I brought up 3.2
> I've attached a screen capture which shows a project with no code change and the same tests ran,
> but it shows a marked drop in coverage that corresponds with when I updated to 3.2
> I did the update late night on 11 August, 2012, This would have been early morning UTC on the 12th.
> Again, I asked if anyone else has seen this behavior?
> Also, what can I do to better analyze why the drop?
>
> Thanks for any suggestions.
>
> Doug
>
> On Wed, Aug 15, 2012 at 08:23:11AM -0600, Doug Beattie wrote:
> > Patroklos,
> > Thanks for the response.
> > The drop was noticed when I switched from 3.0.1 to 3.1.1. (I skipped going to 3.1).
> > From what the teams have said they have not changed any code or tests.
> > Metrics for the LOC/branches, etc. and tests appear to be the same.
> > I also wondered about JaCoCo but that wouldn't have necessarily have come into play until 3.2 although they may have switch to using it with the 3.1.1 release.
> > I'll check into this more.
> > BTW, if the tests are the same why would JaCoCo give different coverage results than Cobetura?
> > If they are different which one was more correct? (Just wondering.)
> >
> > Doug
> >
> >
> > On Wed, Aug 15, 2012 at 10:10:29AM +0300, Papapetrou P.Patroklos wrote:
> > > Hi Doug
> > > The only reason I can think that your code coverage has been lowered when
> > > analysing exactly the same code base is the change of coverage tool. Since
> > > 3.2 the default code coverage tool of Sonar is Jacoco so this might be the
> > > reason but I can't understand why this happened in 3.1.1. Are you sure you
> > > haven't changed anything?
> > > Regards
> > > Patroklos
> > >
> > > 2012/8/14 Doug Beattie <[hidden email]>
> > >
> > > > I have had a few teams approach me lately about coverage numbers on their
> > > > projects being lower.
> > > > Without any code or tests ran being changed, just re-builds/re-analysis,
> > > > the coverage percent has lowered.
> > > > This was noticed from going from 3.0.1 to 3.1.1 and still in the 3.2
> > > > release.
> > > > I have not seen any e-mail on the list about this specifically.
> > > > Has anyone else noticed this?
> > > >
> > > > Doug
> > > > --
> > > >
> > > > ---------------------------------------------------------------------
> > > > To unsubscribe from this list, please visit:
> > > >
> > > >     http://xircles.codehaus.org/manage_email
> > > >
> > > >
> > > >
> >
> > --
> >
> > ---------------------------------------------------------------------
> > To unsubscribe from this list, please visit:
> >
> >     http://xircles.codehaus.org/manage_email
> >
> >
>
> --
>
> ---------------------------------------------------------------------
> To unsubscribe from this list, please visit:
>
>     http://xircles.codehaus.org/manage_email
>
>

--


---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email


dbb
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Coverage seems to have dropped since 3.0.1

dbb
Thank you David for the reply.

Have you seen a difference then between what Cobertura and JaCoCo would report?
We are seeing some major percentage differences, up to 5 percent, on large projects
that have 7-8 thousand lines of code.
I would think that both tools should provide fairly close percentages.

Doug

On Fri, Aug 17, 2012 at 05:05:45PM +0200, David Racodon wrote:

> Hi Doug,
>
> There could be two different explanations:
>
>    1. As Patroklos said, since Sonar 3.2, the default code coverage tool is
>    JaCoCo. So, if no coverage tool was explicitly defined through the
>    'sonar.core.codeCoveragePlugin' property, a switch between Cobertura and
>    JaCoCo has happened. It would explain the differences.
>    2. If the code coverage tool was previously JaCoCo, the code coverage
>    can be slighty different as well because JaCoCo was upgraded from version
>    0.5.6 to 0.5.8. See
>    http://docs.codehaus.org/display/SONAR/Plugin+Versions+Matrix.
>
> Regards,
> *
> *
> *David RACODON | **SonarSource
> **Senior Consultant*
> http://sonarsource.com
>
>
>
> On 17 August 2012 16:32, Doug Beattie <[hidden email]> wrote:
>
> > Forgot to attached the screen shot. Here it is.
> >
> > On Fri, Aug 17, 2012 at 08:29:54AM -0600, Doug Beattie wrote:
> > > It appears the coverage percent drop may have started with 3.2 and not
> > 3.1.
> > > Those reporting to me have all been giving dates that correspond with
> > when I brought up 3.2
> > > I've attached a screen capture which shows a project with no code change
> > and the same tests ran,
> > > but it shows a marked drop in coverage that corresponds with when I
> > updated to 3.2
> > > I did the update late night on 11 August, 2012, This would have been
> > early morning UTC on the 12th.
> > > Again, I asked if anyone else has seen this behavior?
> > > Also, what can I do to better analyze why the drop?
> > >
> > > Thanks for any suggestions.
> > >
> > > Doug
> > >
> > > On Wed, Aug 15, 2012 at 08:23:11AM -0600, Doug Beattie wrote:
> > > > Patroklos,
> > > > Thanks for the response.
> > > > The drop was noticed when I switched from 3.0.1 to 3.1.1. (I skipped
> > going to 3.1).
> > > > From what the teams have said they have not changed any code or tests.
> > > > Metrics for the LOC/branches, etc. and tests appear to be the same.
> > > > I also wondered about JaCoCo but that wouldn't have necessarily have
> > come into play until 3.2 although they may have switch to using it with the
> > 3.1.1 release.
> > > > I'll check into this more.
> > > > BTW, if the tests are the same why would JaCoCo give different
> > coverage results than Cobetura?
> > > > If they are different which one was more correct? (Just wondering.)
> > > >
> > > > Doug
> > > >
> > > >
> > > > On Wed, Aug 15, 2012 at 10:10:29AM +0300, Papapetrou P.Patroklos wrote:
> > > > > Hi Doug
> > > > > The only reason I can think that your code coverage has been lowered
> > when
> > > > > analysing exactly the same code base is the change of coverage tool.
> > Since
> > > > > 3.2 the default code coverage tool of Sonar is Jacoco so this might
> > be the
> > > > > reason but I can't understand why this happened in 3.1.1. Are you
> > sure you
> > > > > haven't changed anything?
> > > > > Regards
> > > > > Patroklos
> > > > >
> > > > > 2012/8/14 Doug Beattie <[hidden email]>
> > > > >
> > > > > > I have had a few teams approach me lately about coverage numbers
> > on their
> > > > > > projects being lower.
> > > > > > Without any code or tests ran being changed, just
> > re-builds/re-analysis,
> > > > > > the coverage percent has lowered.
> > > > > > This was noticed from going from 3.0.1 to 3.1.1 and still in the
> > 3.2
> > > > > > release.
> > > > > > I have not seen any e-mail on the list about this specifically.
> > > > > > Has anyone else noticed this?
> > > > > >
> > > > > > Doug
> > > > > > --
> > > > > >
> > > > > >
> > ---------------------------------------------------------------------
> > > > > > To unsubscribe from this list, please visit:
> > > > > >
> > > > > >     http://xircles.codehaus.org/manage_email
> > > > > >
> > > > > >
> > > > > >
> > > >
> > > > --
> > > >
> > > > ---------------------------------------------------------------------
> > > > To unsubscribe from this list, please visit:
> > > >
> > > >     http://xircles.codehaus.org/manage_email
> > > >
> > > >
> > >
> > > --
> > >
> > > ---------------------------------------------------------------------
> > > To unsubscribe from this list, please visit:
> > >
> > >     http://xircles.codehaus.org/manage_email
> > >
> > >
> >
> > --
> >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe from this list, please visit:
> >
> >     http://xircles.codehaus.org/manage_email
> >
> >

--

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email


Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Coverage seems to have dropped since 3.0.1

David Racodon-2
Hi Doug,

Each tool computes the code coverage a different way, so, yes, there could be some differences.

Regards,

David RACODON | SonarSource
Senior Consultant



On 17 August 2012 19:48, Doug Beattie <[hidden email]> wrote:
Thank you David for the reply.

Have you seen a difference then between what Cobertura and JaCoCo would report?
We are seeing some major percentage differences, up to 5 percent, on large projects
that have 7-8 thousand lines of code.
I would think that both tools should provide fairly close percentages.

Doug

On Fri, Aug 17, 2012 at 05:05:45PM +0200, David Racodon wrote:
> Hi Doug,
>
> There could be two different explanations:
>
>    1. As Patroklos said, since Sonar 3.2, the default code coverage tool is
>    JaCoCo. So, if no coverage tool was explicitly defined through the
>    'sonar.core.codeCoveragePlugin' property, a switch between Cobertura and
>    JaCoCo has happened. It would explain the differences.
>    2. If the code coverage tool was previously JaCoCo, the code coverage
>    can be slighty different as well because JaCoCo was upgraded from version
>    0.5.6 to 0.5.8. See
>    http://docs.codehaus.org/display/SONAR/Plugin+Versions+Matrix.
>
> Regards,
> *
> *
> *David RACODON | **SonarSource
> **Senior Consultant*
> http://sonarsource.com
>
>
>
> On 17 August 2012 16:32, Doug Beattie <[hidden email]> wrote:
>
> > Forgot to attached the screen shot. Here it is.
> >
> > On Fri, Aug 17, 2012 at 08:29:54AM -0600, Doug Beattie wrote:
> > > It appears the coverage percent drop may have started with 3.2 and not
> > 3.1.
> > > Those reporting to me have all been giving dates that correspond with
> > when I brought up 3.2
> > > I've attached a screen capture which shows a project with no code change
> > and the same tests ran,
> > > but it shows a marked drop in coverage that corresponds with when I
> > updated to 3.2
> > > I did the update late night on 11 August, 2012, This would have been
> > early morning UTC on the 12th.
> > > Again, I asked if anyone else has seen this behavior?
> > > Also, what can I do to better analyze why the drop?
> > >
> > > Thanks for any suggestions.
> > >
> > > Doug
> > >
> > > On Wed, Aug 15, 2012 at 08:23:11AM -0600, Doug Beattie wrote:
> > > > Patroklos,
> > > > Thanks for the response.
> > > > The drop was noticed when I switched from 3.0.1 to 3.1.1. (I skipped
> > going to 3.1).
> > > > From what the teams have said they have not changed any code or tests.
> > > > Metrics for the LOC/branches, etc. and tests appear to be the same.
> > > > I also wondered about JaCoCo but that wouldn't have necessarily have
> > come into play until 3.2 although they may have switch to using it with the
> > 3.1.1 release.
> > > > I'll check into this more.
> > > > BTW, if the tests are the same why would JaCoCo give different
> > coverage results than Cobetura?
> > > > If they are different which one was more correct? (Just wondering.)
> > > >
> > > > Doug
> > > >
> > > >
> > > > On Wed, Aug 15, 2012 at 10:10:29AM +0300, Papapetrou P.Patroklos wrote:
> > > > > Hi Doug
> > > > > The only reason I can think that your code coverage has been lowered
> > when
> > > > > analysing exactly the same code base is the change of coverage tool.
> > Since
> > > > > 3.2 the default code coverage tool of Sonar is Jacoco so this might
> > be the
> > > > > reason but I can't understand why this happened in 3.1.1. Are you
> > sure you
> > > > > haven't changed anything?
> > > > > Regards
> > > > > Patroklos
> > > > >
> > > > > 2012/8/14 Doug Beattie <[hidden email]>
> > > > >
> > > > > > I have had a few teams approach me lately about coverage numbers
> > on their
> > > > > > projects being lower.
> > > > > > Without any code or tests ran being changed, just
> > re-builds/re-analysis,
> > > > > > the coverage percent has lowered.
> > > > > > This was noticed from going from 3.0.1 to 3.1.1 and still in the
> > 3.2
> > > > > > release.
> > > > > > I have not seen any e-mail on the list about this specifically.
> > > > > > Has anyone else noticed this?
> > > > > >
> > > > > > Doug
> > > > > > --
> > > > > >
> > > > > >
> > ---------------------------------------------------------------------
> > > > > > To unsubscribe from this list, please visit:
> > > > > >
> > > > > >     http://xircles.codehaus.org/manage_email
> > > > > >
> > > > > >
> > > > > >
> > > >
> > > > --
> > > >
> > > > ---------------------------------------------------------------------
> > > > To unsubscribe from this list, please visit:
> > > >
> > > >     http://xircles.codehaus.org/manage_email
> > > >
> > > >
> > >
> > > --
> > >
> > > ---------------------------------------------------------------------
> > > To unsubscribe from this list, please visit:
> > >
> > >     http://xircles.codehaus.org/manage_email
> > >
> > >
> >
> > --
> >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe from this list, please visit:
> >
> >     http://xircles.codehaus.org/manage_email
> >
> >

--

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email



Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Coverage seems to have dropped since 3.0.1

David Racodon-2
Hi Doug,

As you encounter pretty big differences, feel free to send us of one of your projects, so we can investigate the reasons of the gap

Regards,

David RACODON | SonarSource
Senior Consultant



On 20 August 2012 09:19, David Racodon <[hidden email]> wrote:
Hi Doug,

Each tool computes the code coverage a different way, so, yes, there could be some differences.

Regards,

David RACODON | SonarSource
Senior Consultant



On 17 August 2012 19:48, Doug Beattie <[hidden email]> wrote:
Thank you David for the reply.

Have you seen a difference then between what Cobertura and JaCoCo would report?
We are seeing some major percentage differences, up to 5 percent, on large projects
that have 7-8 thousand lines of code.
I would think that both tools should provide fairly close percentages.

Doug

On Fri, Aug 17, 2012 at 05:05:45PM +0200, David Racodon wrote:
> Hi Doug,
>
> There could be two different explanations:
>
>    1. As Patroklos said, since Sonar 3.2, the default code coverage tool is
>    JaCoCo. So, if no coverage tool was explicitly defined through the
>    'sonar.core.codeCoveragePlugin' property, a switch between Cobertura and
>    JaCoCo has happened. It would explain the differences.
>    2. If the code coverage tool was previously JaCoCo, the code coverage
>    can be slighty different as well because JaCoCo was upgraded from version
>    0.5.6 to 0.5.8. See
>    http://docs.codehaus.org/display/SONAR/Plugin+Versions+Matrix.
>
> Regards,
> *
> *
> *David RACODON | **SonarSource
> **Senior Consultant*
> http://sonarsource.com
>
>
>
> On 17 August 2012 16:32, Doug Beattie <[hidden email]> wrote:
>
> > Forgot to attached the screen shot. Here it is.
> >
> > On Fri, Aug 17, 2012 at 08:29:54AM -0600, Doug Beattie wrote:
> > > It appears the coverage percent drop may have started with 3.2 and not
> > 3.1.
> > > Those reporting to me have all been giving dates that correspond with
> > when I brought up 3.2
> > > I've attached a screen capture which shows a project with no code change
> > and the same tests ran,
> > > but it shows a marked drop in coverage that corresponds with when I
> > updated to 3.2
> > > I did the update late night on 11 August, 2012, This would have been
> > early morning UTC on the 12th.
> > > Again, I asked if anyone else has seen this behavior?
> > > Also, what can I do to better analyze why the drop?
> > >
> > > Thanks for any suggestions.
> > >
> > > Doug
> > >
> > > On Wed, Aug 15, 2012 at 08:23:11AM -0600, Doug Beattie wrote:
> > > > Patroklos,
> > > > Thanks for the response.
> > > > The drop was noticed when I switched from 3.0.1 to 3.1.1. (I skipped
> > going to 3.1).
> > > > From what the teams have said they have not changed any code or tests.
> > > > Metrics for the LOC/branches, etc. and tests appear to be the same.
> > > > I also wondered about JaCoCo but that wouldn't have necessarily have
> > come into play until 3.2 although they may have switch to using it with the
> > 3.1.1 release.
> > > > I'll check into this more.
> > > > BTW, if the tests are the same why would JaCoCo give different
> > coverage results than Cobetura?
> > > > If they are different which one was more correct? (Just wondering.)
> > > >
> > > > Doug
> > > >
> > > >
> > > > On Wed, Aug 15, 2012 at 10:10:29AM +0300, Papapetrou P.Patroklos wrote:
> > > > > Hi Doug
> > > > > The only reason I can think that your code coverage has been lowered
> > when
> > > > > analysing exactly the same code base is the change of coverage tool.
> > Since
> > > > > 3.2 the default code coverage tool of Sonar is Jacoco so this might
> > be the
> > > > > reason but I can't understand why this happened in 3.1.1. Are you
> > sure you
> > > > > haven't changed anything?
> > > > > Regards
> > > > > Patroklos
> > > > >
> > > > > 2012/8/14 Doug Beattie <[hidden email]>
> > > > >
> > > > > > I have had a few teams approach me lately about coverage numbers
> > on their
> > > > > > projects being lower.
> > > > > > Without any code or tests ran being changed, just
> > re-builds/re-analysis,
> > > > > > the coverage percent has lowered.
> > > > > > This was noticed from going from 3.0.1 to 3.1.1 and still in the
> > 3.2
> > > > > > release.
> > > > > > I have not seen any e-mail on the list about this specifically.
> > > > > > Has anyone else noticed this?
> > > > > >
> > > > > > Doug
> > > > > > --
> > > > > >
> > > > > >
> > ---------------------------------------------------------------------
> > > > > > To unsubscribe from this list, please visit:
> > > > > >
> > > > > >     http://xircles.codehaus.org/manage_email
> > > > > >
> > > > > >
> > > > > >
> > > >
> > > > --
> > > >
> > > > ---------------------------------------------------------------------
> > > > To unsubscribe from this list, please visit:
> > > >
> > > >     http://xircles.codehaus.org/manage_email
> > > >
> > > >
> > >
> > > --
> > >
> > > ---------------------------------------------------------------------
> > > To unsubscribe from this list, please visit:
> > >
> > >     http://xircles.codehaus.org/manage_email
> > >
> > >
> >
> > --
> >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe from this list, please visit:
> >
> >     http://xircles.codehaus.org/manage_email
> >
> >

--

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email




Loading...