Failing Test under Windows in Stealth-Mode

[hdockter]

Hi,

we had a regression bug that Gradle always returns exit value 0, even  
in case of an error. Since September 9 there are unit tests failing  
specifically under Windows (none of the Gradle developers is using  
this OS). Our windows CI build did not signal this failing unit tests  
because of the above bug. Fortunately those were bugs just in the  
unit tests and not in our production code.

The real problem is not the bug, but the fact that none of our tests  
did detect this. I'm going to add a couple of integration tests to  
check process return value behaviour.

Adam, I guess we have to draw lots for who of us must switch to  
windows ;)

- Hans

--
Hans Dockter
Gradle Project lead
http://www.gradle.org





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

    http://xircles.codehaus.org/manage_email

[Helmut Denk]

i noticed that because i regularily build and install the SNAPSHOT on
my windows system (xp sp2).

i did a -Dskip.test and -Dskip.integTest to overcome this.

i did not mention it in the forums as i am too new
in gradle. may be next time i will have the guts to do so ;-)

IMO the ci-build is exectly the right tool to avoid that
situation ...

have a nice day.

hdockter wrote:

Hi,

we had a regression bug that Gradle always returns exit value 0, even  
in case of an error. Since September 9 there are unit tests failing  
specifically under Windows (none of the Gradle developers is using  
this OS). Our windows CI build did not signal this failing unit tests  
because of the above bug. Fortunately those were bugs just in the  
unit tests and not in our production code.

The real problem is not the bug, but the fact that none of our tests  
did detect this. I'm going to add a couple of integration tests to  
check process return value behaviour.

Adam, I guess we have to draw lots for who of us must switch to  
windows ;)

- Hans

--
Hans Dockter
Gradle Project lead
http://www.gradle.org





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

    http://xircles.codehaus.org/manage_email

[hdockter]

On Sep 18, 2008, at 12:55 PM, Helmut Denk wrote:

>
> i noticed that because i regularily build and install the SNAPSHOT on
> my windows system (xp sp2).
>
> i did a -Dskip.test and -Dskip.integTest to overcome this.
>
> i did not mention it in the forums as i am too new
> in gradle. may be next time i will have the guts to do so ;-)
>
> IMO the ci-build is exectly the right tool to avoid that
> situation ...

Definitely. It was just a joke that the developers have to switch.  
And the problem is fixed now. This means:

- any failing test will now lead again to a message from our CI  
windows/linux build.
- there are no failing tests under windows any longer.

- Hans

--
Hans Dockter
Gradle Project lead
http://www.gradle.org





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

    http://xircles.codehaus.org/manage_email

[Helmut Denk]

i still get 1 failing test for the trunk (24. september):

*** schnipp ***
junit.framework.AssertionFailedError: at org.gradle.groovy.scripts.DefaultScriptHandlerTest.testLoadFromCacheWithStaleCache(DefaultScriptHandlerTest.java:108)
*** schnapp ***

java.version 1.6.0_07
os.name Windows XP
os.version 5.1

to be honest, i do'nt know, what 'Stealth-Mode' is.

have a nice day

[hdockter]

On Sep 24, 2008, at 8:57 AM, Helmut Denk wrote:

>
> i still get 1 failing test for the trunk (24. september):
>
> *** schnipp ***
> junit.framework.AssertionFailedError: at
> org.gradle.groovy.scripts.DefaultScriptHandlerTest.testLoadFromCacheWi
> thStaleCache(DefaultScriptHandlerTest.java:108)
> *** schnapp ***
>
> java.version 1.6.0_07
> os.name Windows XP
> os.version 5.1

This is strange. On our windows CI build (http://
teamcity.jetbrains.com/repository/download/bt41/14091:id/TEST-
org.gradle.groovy.scripts.DefaultScriptHandlerTest.xml) as well as  
under my virtual windows machine on my Mac this test is succeeding.
The CI machine is having the same environment as you have.

>
> to be honest, i do'nt know, what 'Stealth-Mode' is.

Just figurative language. It just means they have failed yet nobody  
got notified. This problem is fixed.

- Hans

--
Hans Dockter
Gradle Project lead
http://www.gradle.org





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

    http://xircles.codehaus.org/manage_email

[Helmut Denk]

hdockter wrote:

This is strange. On our windows CI build (http://
teamcity.jetbrains.com/repository/download/bt41/14091:id/TEST-
org.gradle.groovy.scripts.DefaultScriptHandlerTest.xml) as well as  
under my virtual windows machine on my Mac this test is succeeding.
The CI machine is having the same environment as you have.

i did a clean-checkout and deleted .grade dir in the user.home
to exclude sideeffects by previous build ... same error.
all other units-tests are ok and integTests run ok too.
ca'nt do more about it at this moment ...

[Adam Murdoch-2]

Hi,

Which filesystem type are you using? Some of them are a bit particular
about file timestamps, which is what this test uses.

I've checked in a change to the test, if you could give it a try. If the
timestamps are the problem, the test will still fail, but on a different
assertion. And then we can figure out how to fix it.


Adam

Helmut Denk wrote:

> i still get 1 failing test for the trunk (24. september):
>
> *** schnipp ***
> junit.framework.AssertionFailedError: at
> org.gradle.groovy.scripts.DefaultScriptHandlerTest.testLoadFromCacheWithStaleCache(DefaultScriptHandlerTest.java:108)
> *** schnapp ***
>
> java.version 1.6.0_07
> os.name Windows XP
> os.version 5.1
>
> to be honest, i do'nt know, what 'Stealth-Mode' is.
>
> have a nice day
>
>
>  

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

    http://xircles.codehaus.org/manage_email

[Helmut Denk]

hello adam,

Adam Murdoch-2 wrote:

Which filesystem type are you using?

NTFS nothing special about it afaik

Adam Murdoch-2 wrote:

I've checked in a change to the test, if you could give it a try. If the
timestamps are the problem, the test will still fail, but on a different
assertion. And then we can figure out how to fix it.

ok. done it. i changed your statement a little bit:

>>assertTrue("adams new assertion failed", scriptCacheDir.lastModified() < 100000);

this is what i get:

junit.framework.AssertionFailedError: adams new assertion failed
at org.gradle.groovy.scripts.DefaultScriptHandlerTest.testLoadFromCacheWithStaleCache(DefaultScriptHandlerTest.java:108)

i hope that helps ...

[Adam Murdoch-2]

Hi,

I've checked in what is hopefully a fix for this, if you could try it.


Helmut Denk wrote:

hello adam,


Adam Murdoch-2 wrote:
  

Which filesystem type are you using? 

    

NTFS nothing special about it afaik


Adam Murdoch-2 wrote:
  

I've checked in a change to the test, if you could give it a try. If the 
timestamps are the problem, the test will still fail, but on a different 
assertion. And then we can figure out how to fix it.

    

ok. done it. i changed your statement a little bit:

  

assertTrue("adams new assertion failed", scriptCacheDir.lastModified() <
      

100000);

this is what i get:

junit.framework.AssertionFailedError: adams new assertion failed
at
org.gradle.groovy.scripts.DefaultScriptHandlerTest.testLoadFromCacheWithStaleCache(DefaultScriptHandlerTest.java:108)

i hope that helps ...


  

--------------------------------------------------------------------- To unsubscribe from this list, please visit: http://xircles.codehaus.org/manage_email

[Helmut Denk]

the test runs ok now.

Adam Murdoch-2 wrote:

Hi,

I've checked in what is hopefully a fix for this, if you could try it.