You are viewing a plain text version of this content. The canonical link for it is here.
Posted to cactus-dev@jakarta.apache.org by Curt Arnold <ca...@apache.org> on 2005/02/23 06:55:39 UTC
Recent Gump breakage
There have been recent changes to log4j that have broken the Gump build
of jakarta-hivemind. The start of yourt Gump build failures coincide
with the log4j changes and the breaking of hivemind. Do you know if
our recent changes were the cause of your Gump failures too?
---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org
Re: Recent Gump breakage
Posted by Curt Arnold <ca...@apache.org>.
On Feb 23, 2005, at 3:01 AM, Stefan Bodewig wrote:
> On Tue, 22 Feb 2005, Curt Arnold <ca...@apache.org> wrote:
>
>> There have been recent changes to log4j that have broken the Gump
>> build of jakarta-hivemind.
>
> I guess you didn't mean to send this to cactus-dev
>
I did. The recent breakages of HiveMind were known to be caused by
changes in log4j. Several of the Cactus projects
(jakarta-cactus-sample-servlet-12, jakarta-cactus-sample-servlet-13,
jakarta-cactus-sample-jetty-13) broke at the same time with no obvious
change in Cactus which suggests that they too may have been broken by
the changes in log4j.
> I took some time to look into the hivemind test failures yesterday.
> The tests try to set up a new appender which is supposed to capture
> the log output, using
>
> Logger logger = LogManager.getLogger(loggerName);
>
> logger.removeAllAppenders();
>
> _interceptedLoggerName = loggerName;
> _appender = new StoreAppender();
>
> logger.setLevel(Level.DEBUG);
> logger.setAdditivity(false);
> logger.addAppender(_appender);
>
> and I think, lines looking like
>
> [junit] 1 [main] ERROR org.apache.hivemind.test.StoreAppender -
> Attempted to log with inactive appender named [null].
>
> have not been there before the tests failed (I could be wrong,
> though).
>
> Maybe this gives anybody a clue.
>
The first phase of the HiveMind failures were reports were compilation
failures due to new methods added to the Appender interface. After
that interface was reverted, the "inactive appender" messages started.
As the code stands at this instance, I think that you are newly
required to call _appender.activate() before adding an appender to a
logger. I guess I'm going to have to attempt to set up a HiveMind
build to make sure that I fix all the problems this time.
---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org
Re: Recent Gump breakage
Posted by Stefan Bodewig <bo...@apache.org>.
On Tue, 22 Feb 2005, Curt Arnold <ca...@apache.org> wrote:
> There have been recent changes to log4j that have broken the Gump
> build of jakarta-hivemind.
I guess you didn't mean to send this to cactus-dev
I took some time to look into the hivemind test failures yesterday.
The tests try to set up a new appender which is supposed to capture
the log output, using
Logger logger = LogManager.getLogger(loggerName);
logger.removeAllAppenders();
_interceptedLoggerName = loggerName;
_appender = new StoreAppender();
logger.setLevel(Level.DEBUG);
logger.setAdditivity(false);
logger.addAppender(_appender);
and I think, lines looking like
[junit] 1 [main] ERROR org.apache.hivemind.test.StoreAppender - Attempted to log with inactive appender named [null].
have not been there before the tests failed (I could be wrong,
though).
Maybe this gives anybody a clue.
Stefan
---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org
Re: Recent Gump breakage
Posted by Stefan Bodewig <bo...@apache.org>.
On Thu, 24 Feb 2005, Curt Arnold <ca...@apache.org> wrote:
> We have analyzed the Gump breakage of Hivemind and have been able to
> suppress the problem.
Cool.
> However it didn't clean up the problem with the Cactus Gump runs.
The three Cactus projects that fail, have been failing for a long
time. I don't think it is related to log4j at all. That they started
to fail around the time of the log4j changes is pure coincidence IMHO.
Before that they didn't even build because of a failing pre-req
(mx4j-tools).
Stefan
---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org
RE: Recent Gump breakage
Posted by Vincent Massol <vm...@pivolis.com>.
Hi Curt,
Thanks for all your analysis. The problem is that Cactus committers have not
touched Gump for a very long time now. I used to use it but then I've lost
faith somewhere along the line (it was taking me more time to fix Gump and
Co than to fix Cactus and most of the time the problems were either because
of Gump or because of other projects and I didn't have the bandwidth to help
fix that - Mind you I've done it for several years :-)). For the record, I
do believe Gump is an extremely useful tool.
I'm hoping that some Cactus committer show up and take the lead on fixing
the Cactus Gump descriptors. Cactus is building well on my machine and I
believe it's only a question of fixing the Cactus Gump descriptor...
Thanks
-Vincent
> -----Original Message-----
> From: Curt Arnold [mailto:carnold@apache.org]
> Sent: jeudi 24 février 2005 22:36
> To: Cactus Developers List
> Subject: Re: Recent Gump breakage
>
> We have analyzed the Gump breakage of Hivemind and have been able to
> suppress the problem. However it didn't clean up the problem with the
> Cactus Gump runs. I made a half-hearted effort to set up a build
> environment for Cactus but abandoned it. It would be really helpful if
> someone could check that log4j was responsible for the breakage but
> running any of the affected tests with both a pre-17 Feb log4j and a
> post-17 Feb log4j (you could just use the ones from Gump). If so, it
> would be extremely lovely if you can try to figure out the nature of
> the change in behavior that caused Cactus to break.
>
> I may try to do some Gump shenanigans to have it help.
>
>
> On Feb 22, 2005, at 11:55 PM, Curt Arnold wrote:
>
> > There have been recent changes to log4j that have broken the Gump
> > build of jakarta-hivemind. The start of yourt Gump build failures
> > coincide with the log4j changes and the breaking of hivemind. Do you
> > know if our recent changes were the cause of your Gump failures too?
> >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
> > For additional commands, e-mail: cactus-dev-help@jakarta.apache.org
> >
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: cactus-dev-help@jakarta.apache.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org
Re: Recent Gump breakage
Posted by Curt Arnold <ca...@apache.org>.
We have analyzed the Gump breakage of Hivemind and have been able to
suppress the problem. However it didn't clean up the problem with the
Cactus Gump runs. I made a half-hearted effort to set up a build
environment for Cactus but abandoned it. It would be really helpful if
someone could check that log4j was responsible for the breakage but
running any of the affected tests with both a pre-17 Feb log4j and a
post-17 Feb log4j (you could just use the ones from Gump). If so, it
would be extremely lovely if you can try to figure out the nature of
the change in behavior that caused Cactus to break.
I may try to do some Gump shenanigans to have it help.
On Feb 22, 2005, at 11:55 PM, Curt Arnold wrote:
> There have been recent changes to log4j that have broken the Gump
> build of jakarta-hivemind. The start of yourt Gump build failures
> coincide with the log4j changes and the breaking of hivemind. Do you
> know if our recent changes were the cause of your Gump failures too?
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: cactus-dev-help@jakarta.apache.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org
RE: Recent Gump breakage
Posted by Vincent Massol <vm...@pivolis.com>.
Hi Curt,
Sorry but I don't know. I've not been following Gump for ages now. It would
be good if some other Cactus committers were interested in working on the
Cactus Gump build... (this is a call for help! :-)).
Thanks
-Vincent
PS: I'm going on holiday today for 2 weeks...
> -----Original Message-----
> From: Curt Arnold [mailto:carnold@apache.org]
> Sent: mercredi 23 février 2005 06:56
> To: cactus-dev@jakarta.apache.org
> Subject: Recent Gump breakage
>
> There have been recent changes to log4j that have broken the Gump build
> of jakarta-hivemind. The start of yourt Gump build failures coincide
> with the log4j changes and the breaking of hivemind. Do you know if
> our recent changes were the cause of your Gump failures too?
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: cactus-dev-help@jakarta.apache.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org