Re: Thoughts on MNG-5626 and the need for a log file

2014-05-14 Thread Bernd Eckenfels
Sorry I was not clear, what I was trying to say is: maven has logging
built in, if you need a logfile you can pass the -l switch or
re-configure slf4j. It also has some control over the loglevel (-X).
Most users dont need it because they (think CI Server) just catch
stdout.

http://maven.apache.org/maven-logging.html

Gruss
Bernd

Am Mon, 5 May 2014 15:14:33 -0500 schrieb Paul
Benedict pbened...@apache.org:

 -X is debug/stack traces, right? That's great when you want a
 slosh-load of debug information, but that's only useful when you're
 looking for a problem. There's no recourse when problems oddball
 conditions occur with -X not enabled. My point is Maven and Maven
 plugins should be able to log those abnormal conditions in the course
 of a normal build. We shouldn't have to dump the world -- especially
 when the problem is not always repeatable.
 
 
 
 
 Cheers,
 Paul
 
 
 On Mon, May 5, 2014 at 2:38 PM, Bernd Eckenfels
 e...@zusammenkunft.netwrote:
 
  Am Mon, 5 May 2014 14:17:50 -0500
  schrieb Paul Benedict pbened...@apache.org:
 
   What should be logged? Warnings. I actually would propose
   exceptions too because them on the screen only help if you have
   logging like a Hudson instance to view them.
 
  Uh, what is wrong with mvn -X -l?
 
  Gruss
  Bernd
 
  -
  To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
  For additional commands, e-mail: dev-h...@maven.apache.org
 
 
 

-
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org



Thoughts on MNG-5626 and the need for a log file

2014-05-05 Thread Paul Benedict
One thing that I like about Eclipse is that it contains a log file to
capture the unexpected warning or error. These warnings or errors may not
kill the program but at least I can peer inside to see what's going on.

With regard to MNG-5626, it makes me wonder should Maven have a default
logging location. There are situations that shouldn't kill a build (like
negative build times) but are extraordinary enough that they should be
dumped to a log file for studying. I think plugins should have the ability
to do such things for the sake of diagnosing out unfavorable
conditions/bugs in the code.

BTW, this is a different feature than debug info and stack traces. I don't
want to bug the user with more on their screen. I just want normal builds
to run like they do except introduce a warning log.

Thoughts?

Cheers,
Paul


Re: Thoughts on MNG-5626 and the need for a log file

2014-05-05 Thread Michael Osipov

Am 2014-05-05 16:20, schrieb Paul Benedict:

One thing that I like about Eclipse is that it contains a log file to
capture the unexpected warning or error. These warnings or errors may not
kill the program but at least I can peer inside to see what's going on.

With regard to MNG-5626, it makes me wonder should Maven have a default
logging location. There are situations that shouldn't kill a build (like
negative build times) but are extraordinary enough that they should be
dumped to a log file for studying. I think plugins should have the ability
to do such things for the sake of diagnosing out unfavorable
conditions/bugs in the code.

BTW, this is a different feature than debug info and stack traces. I don't
want to bug the user with more on their screen. I just want normal builds
to run like they do except introduce a warning log.


Paul,

how would a log file help to solve the above mentioned problem 
(MNG-5626). I guess Logback relies on currentMillis too.


Moreover, what should be logged and to what extent?

I do think, it's worth investigating but quite hard to decide what 
should be printed to such a file. Using SLF4J markers and a distinct 
Logback configuration may be a good help.


Michael


-
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org



Re: Thoughts on MNG-5626 and the need for a log file

2014-05-05 Thread Paul Benedict
In the instance of MNG-5256, the log file could dump the values that go
into calculating a negative build time. I think that would be very
insightful to trace the root cause.

What should be logged? Warnings. I actually would propose exceptions too
because them on the screen only help if you have logging like a Hudson
instance to view them.

Paul


Cheers,
Paul


On Mon, May 5, 2014 at 1:49 PM, Michael Osipov micha...@apache.org wrote:

 Am 2014-05-05 16:20, schrieb Paul Benedict:

  One thing that I like about Eclipse is that it contains a log file to
 capture the unexpected warning or error. These warnings or errors may not
 kill the program but at least I can peer inside to see what's going on.

 With regard to MNG-5626, it makes me wonder should Maven have a default
 logging location. There are situations that shouldn't kill a build (like
 negative build times) but are extraordinary enough that they should be
 dumped to a log file for studying. I think plugins should have the ability
 to do such things for the sake of diagnosing out unfavorable
 conditions/bugs in the code.

 BTW, this is a different feature than debug info and stack traces. I don't
 want to bug the user with more on their screen. I just want normal builds
 to run like they do except introduce a warning log.


 Paul,

 how would a log file help to solve the above mentioned problem (MNG-5626).
 I guess Logback relies on currentMillis too.

 Moreover, what should be logged and to what extent?

 I do think, it's worth investigating but quite hard to decide what should
 be printed to such a file. Using SLF4J markers and a distinct Logback
 configuration may be a good help.

 Michael


 -
 To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
 For additional commands, e-mail: dev-h...@maven.apache.org




Re: Thoughts on MNG-5626 and the need for a log file

2014-05-05 Thread Bernd Eckenfels
Am Mon, 5 May 2014 14:17:50 -0500
schrieb Paul Benedict pbened...@apache.org:

 What should be logged? Warnings. I actually would propose exceptions
 too because them on the screen only help if you have logging like a
 Hudson instance to view them.

Uh, what is wrong with mvn -X -l?

Gruss
Bernd

-
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org



Re: Thoughts on MNG-5626 and the need for a log file

2014-05-05 Thread Paul Benedict
-X is debug/stack traces, right? That's great when you want a slosh-load of
debug information, but that's only useful when you're looking for a
problem. There's no recourse when problems oddball conditions occur with -X
not enabled. My point is Maven and Maven plugins should be able to log
those abnormal conditions in the course of a normal build. We shouldn't
have to dump the world -- especially when the problem is not always
repeatable.




Cheers,
Paul


On Mon, May 5, 2014 at 2:38 PM, Bernd Eckenfels e...@zusammenkunft.netwrote:

 Am Mon, 5 May 2014 14:17:50 -0500
 schrieb Paul Benedict pbened...@apache.org:

  What should be logged? Warnings. I actually would propose exceptions
  too because them on the screen only help if you have logging like a
  Hudson instance to view them.

 Uh, what is wrong with mvn -X -l?

 Gruss
 Bernd

 -
 To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
 For additional commands, e-mail: dev-h...@maven.apache.org




Re: Thoughts on MNG-5626 and the need for a log file

2014-05-05 Thread Bernd Eckenfels
Sorry I was not clear, what I was trying to say is: maven has logging
built in, if you need a logfile you can pass the -l switch or
re-configure slf4j. It also has some control over the loglevel (-X).
Most users dont need it because they (think CI Server) just catch
stdout.

http://maven.apache.org/maven-logging.html

Gruss
Bernd

Am Mon, 5 May 2014 15:14:33 -0500 schrieb Paul
Benedict pbened...@apache.org:

 -X is debug/stack traces, right? That's great when you want a
 slosh-load of debug information, but that's only useful when you're
 looking for a problem. There's no recourse when problems oddball
 conditions occur with -X not enabled. My point is Maven and Maven
 plugins should be able to log those abnormal conditions in the course
 of a normal build. We shouldn't have to dump the world -- especially
 when the problem is not always repeatable.
 
 
 
 
 Cheers,
 Paul
 
 
 On Mon, May 5, 2014 at 2:38 PM, Bernd Eckenfels
 e...@zusammenkunft.netwrote:
 
  Am Mon, 5 May 2014 14:17:50 -0500
  schrieb Paul Benedict pbened...@apache.org:
 
   What should be logged? Warnings. I actually would propose
   exceptions too because them on the screen only help if you have
   logging like a Hudson instance to view them.
 
  Uh, what is wrong with mvn -X -l?
 
  Gruss
  Bernd
 
  -
  To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
  For additional commands, e-mail: dev-h...@maven.apache.org
 
 
 

-
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org