You are viewing a plain text version of this content. The canonical link for it is here.
Posted to log4j-dev@logging.apache.org by Curt Arnold <ca...@apache.org> on 2004/12/21 09:13:48 UTC
RepositorySelector.detachRepository, jetty-plus, gump and PatternLayout.test14
The most dreaded nightmare of a new committer is breaking the build. I
don't think that I have (and if so only log4j-tests), but I have been
having intermittent problems with PatternLayoutTest.test14 and was
trying to figure out if it was just me.
I was checking Gump and it looks like that log4j-tests haven't been
running since August, see
http://brutus.apache.org/gump/public/logging-log4j/log4j-tests/
index.html.
At least one reason for this failure is because jetty-plus no longer
builds
(http://brutus.apache.org/gump/public/jetty/jetty-plus/index.html)
which appears to be due to detachRepository being added to the
RepositorySelector interface back on 30 March 2004. Don't know why we
were able to build in August.
Was the change to the RepositorySelector an intentional break of
earlier implementations of the interface?
p.s. when PatternLayoutTest.test14 fails for me, the output file is
empty (the witness file isn't). I think it is likely a configuration
problem, but I have seen it pass then fail on the same checkout.
---------------------------------------------------------------------
To unsubscribe, e-mail: log4j-dev-unsubscribe@logging.apache.org
For additional commands, e-mail: log4j-dev-help@logging.apache.org
Re: RepositorySelector.detachRepository, jetty-plus, gump and
PatternLayout.test14
Posted by Ceki Gülcü <ce...@qos.ch>.
At 09:13 AM 12/21/2004, Curt Arnold wrote:
>The most dreaded nightmare of a new committer is breaking the build. I
>don't think that I have (and if so only log4j-tests), but I have been
>having intermittent problems with PatternLayoutTest.test14 and was
>trying to figure out if it was just me.
PatternLayout.14 seems to be special in two respects. First, it contains
spaces for the value of
the "log4j.appender.testAppender.File" property.
The line is
log4j.appender.testAppender.File= output/temp\s\s
where \s represents the space character.
The conversion pattern in PatternLayout.14 uses the %counter conversion
word which is defined by MyPatternLayout. Its source code can be found
under examples/src/pattern/MyPatternLayout.java.
So the problem is probably due to a dependency with examples/ that Jake has
already complained about.
--
Ceki Gülcü
The complete log4j manual: http://qos.ch/log4j/
---------------------------------------------------------------------
To unsubscribe, e-mail: log4j-dev-unsubscribe@logging.apache.org
For additional commands, e-mail: log4j-dev-help@logging.apache.org
Re: RepositorySelector.detachRepository, jetty-plus, gump and
PatternLayout.test14
Posted by Ceki Gülcü <ce...@qos.ch>.
The dependency on jetty for the log4j GUMP project descriptor has been
removed. This should allow log4j-tests to proceed.
I still don't see how building jetty could fail to build do to a change in
log4j.
At 09:54 PM 12/22/2004, Ceki Gülcü wrote:
>As I remember it, running log4j-tests by GUMP started failing when jetty
>became a dependency of log4j-tests.
>
>Your observation about log4j-tests failing due to the addition of
>detachRepository method in RepositorySelector is news to me. Where does
>that come form?
>
>The detachRepository method in RepositorySelector ensures that
>repositories can be recycled as a web-app is recycled, so it is of
>paramount importance.
>
>
>At 09:14 PM 12/22/2004, Curt Arnold wrote:
>>This would be really good to resolve. As it stands, the GUMP automated
>>runs of our unit tests have been offline for several months and there
>>is no consistency on the state of our unit tests. That forces me to
>>first figure out what tests are failing before I made a modification,
>>hack the code so the rest of the tests pass, and then test my mod. If
>>you know that you are some tests for a while, I don't have a problem
>>with the tests being commented out or otherwise bypassed as long as you
>>remember to add them back.
>>
>>So, did we intend to break existing implementations of
>>RepositorySelector by adding the detachRepository method? How can we
>>resolve the issue:
>>
>>Have Jetty declare that it depends on 1.2
>>
>>implement detachRepository knowing that it would be ignored when
>>building on 1.2 (and if so what should it do)
>>
>>Revert RepositorySelector and introduce a new interface for
>>detachRepository
>>
>>
>>On Dec 21, 2004, at 2:13 AM, Curt Arnold wrote:
>>
>>>The most dreaded nightmare of a new committer is breaking the build.
>>>I don't think that I have (and if so only log4j-tests), but I have
>>>been having intermittent problems with PatternLayoutTest.test14 and
>>>was trying to figure out if it was just me.
>>>
>>>I was checking Gump and it looks like that log4j-tests haven't been
>>>running since August, see
>>>http://brutus.apache.org/gump/public/logging-log4j/log4j-tests/ index.html.
>>>
>>>At least one reason for this failure is because jetty-plus no longer
>>>builds
>>>(http://brutus.apache.org/gump/public/jetty/jetty-plus/index.html)
>>>which appears to be due to detachRepository being added to the
>>>RepositorySelector interface back on 30 March 2004. Don't know why we
>>>were able to build in August.
>>>
>>>Was the change to the RepositorySelector an intentional break of
>>>earlier implementations of the interface?
>>>
>>>p.s. when PatternLayoutTest.test14 fails for me, the output file is
>>>empty (the witness file isn't). I think it is likely a configuration
>>>problem, but I have seen it pass then fail on the same checkout.
>>
>>
>>---------------------------------------------------------------------
>>To unsubscribe, e-mail: log4j-dev-unsubscribe@logging.apache.org
>>For additional commands, e-mail: log4j-dev-help@logging.apache.org
>
>--
>Ceki Gülcü
>
> The complete log4j manual: http://qos.ch/log4j/
>
>
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: log4j-dev-unsubscribe@logging.apache.org
>For additional commands, e-mail: log4j-dev-help@logging.apache.org
--
Ceki Gülcü
The complete log4j manual: http://qos.ch/log4j/
---------------------------------------------------------------------
To unsubscribe, e-mail: log4j-dev-unsubscribe@logging.apache.org
For additional commands, e-mail: log4j-dev-help@logging.apache.org
Re: RepositorySelector.detachRepository, jetty-plus, gump and PatternLayout.test14
Posted by Curt Arnold <ca...@apache.org>.
On Dec 22, 2004, at 2:54 PM, Ceki Gülcü wrote:
> Your observation about log4j-tests failing due to the addition of
> detachRepository method in RepositorySelector is news to me. Where
> does that come form?
>
Jetty appear to be failing to compile since it has a class that
implements RepositorySelector but doesn't implement detachRepository.
Since jetty fails to compile, log4j-test is never attempted.
> The detachRepository method in RepositorySelector ensures that
> repositories can be recycled as a web-app is recycled, so it is of
> paramount importance.
>
Can we suggest that Jetty implement detachRepository. If they run
against a 1.2 log4j, the method will never get called.
---------------------------------------------------------------------
To unsubscribe, e-mail: log4j-dev-unsubscribe@logging.apache.org
For additional commands, e-mail: log4j-dev-help@logging.apache.org
Re: RepositorySelector.detachRepository, jetty-plus, gump and
PatternLayout.test14
Posted by Ceki Gülcü <ce...@qos.ch>.
As I remember it, running log4j-tests by GUMP started failing when jetty
became a dependency of log4j-tests.
Your observation about log4j-tests failing due to the addition of
detachRepository method in RepositorySelector is news to me. Where does
that come form?
The detachRepository method in RepositorySelector ensures that repositories
can be recycled as a web-app is recycled, so it is of paramount importance.
At 09:14 PM 12/22/2004, Curt Arnold wrote:
>This would be really good to resolve. As it stands, the GUMP automated
>runs of our unit tests have been offline for several months and there
>is no consistency on the state of our unit tests. That forces me to
>first figure out what tests are failing before I made a modification,
>hack the code so the rest of the tests pass, and then test my mod. If
>you know that you are some tests for a while, I don't have a problem
>with the tests being commented out or otherwise bypassed as long as you
>remember to add them back.
>
>So, did we intend to break existing implementations of
>RepositorySelector by adding the detachRepository method? How can we
>resolve the issue:
>
>Have Jetty declare that it depends on 1.2
>
>implement detachRepository knowing that it would be ignored when
>building on 1.2 (and if so what should it do)
>
>Revert RepositorySelector and introduce a new interface for
>detachRepository
>
>
>On Dec 21, 2004, at 2:13 AM, Curt Arnold wrote:
>
>>The most dreaded nightmare of a new committer is breaking the build.
>>I don't think that I have (and if so only log4j-tests), but I have
>>been having intermittent problems with PatternLayoutTest.test14 and
>>was trying to figure out if it was just me.
>>
>>I was checking Gump and it looks like that log4j-tests haven't been
>>running since August, see
>>http://brutus.apache.org/gump/public/logging-log4j/log4j-tests/ index.html.
>>
>>At least one reason for this failure is because jetty-plus no longer
>>builds
>>(http://brutus.apache.org/gump/public/jetty/jetty-plus/index.html)
>>which appears to be due to detachRepository being added to the
>>RepositorySelector interface back on 30 March 2004. Don't know why we
>>were able to build in August.
>>
>>Was the change to the RepositorySelector an intentional break of
>>earlier implementations of the interface?
>>
>>p.s. when PatternLayoutTest.test14 fails for me, the output file is
>>empty (the witness file isn't). I think it is likely a configuration
>>problem, but I have seen it pass then fail on the same checkout.
>
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: log4j-dev-unsubscribe@logging.apache.org
>For additional commands, e-mail: log4j-dev-help@logging.apache.org
--
Ceki Gülcü
The complete log4j manual: http://qos.ch/log4j/
---------------------------------------------------------------------
To unsubscribe, e-mail: log4j-dev-unsubscribe@logging.apache.org
For additional commands, e-mail: log4j-dev-help@logging.apache.org
Re: RepositorySelector.detachRepository, jetty-plus, gump and PatternLayout.test14
Posted by Curt Arnold <ca...@apache.org>.
This would be really good to resolve. As it stands, the GUMP automated
runs of our unit tests have been offline for several months and there
is no consistency on the state of our unit tests. That forces me to
first figure out what tests are failing before I made a modification,
hack the code so the rest of the tests pass, and then test my mod. If
you know that you are some tests for a while, I don't have a problem
with the tests being commented out or otherwise bypassed as long as you
remember to add them back.
So, did we intend to break existing implementations of
RepositorySelector by adding the detachRepository method? How can we
resolve the issue:
Have Jetty declare that it depends on 1.2
implement detachRepository knowing that it would be ignored when
building on 1.2 (and if so what should it do)
Revert RepositorySelector and introduce a new interface for
detachRepository
On Dec 21, 2004, at 2:13 AM, Curt Arnold wrote:
> The most dreaded nightmare of a new committer is breaking the build.
> I don't think that I have (and if so only log4j-tests), but I have
> been having intermittent problems with PatternLayoutTest.test14 and
> was trying to figure out if it was just me.
>
> I was checking Gump and it looks like that log4j-tests haven't been
> running since August, see
> http://brutus.apache.org/gump/public/logging-log4j/log4j-tests/
> index.html.
>
> At least one reason for this failure is because jetty-plus no longer
> builds
> (http://brutus.apache.org/gump/public/jetty/jetty-plus/index.html)
> which appears to be due to detachRepository being added to the
> RepositorySelector interface back on 30 March 2004. Don't know why we
> were able to build in August.
>
> Was the change to the RepositorySelector an intentional break of
> earlier implementations of the interface?
>
> p.s. when PatternLayoutTest.test14 fails for me, the output file is
> empty (the witness file isn't). I think it is likely a configuration
> problem, but I have seen it pass then fail on the same checkout.
---------------------------------------------------------------------
To unsubscribe, e-mail: log4j-dev-unsubscribe@logging.apache.org
For additional commands, e-mail: log4j-dev-help@logging.apache.org