- logback-user - mailman3.qos.ch

Re: [logback-user] jndi db access in tomcat5.5
by Anton Tagunov 17 Jul '08

17 Jul '08

Hello Diego, [1] > using tomcat5.5, username/password can't be specified when > trying to use a JNDI-shared DataSource > when using pooled data sources the user and > password version of getConnection method can't be used. > I don't know if this is tomcat-specific behaviour. this is indeed an iteresting question; do we have anybody here with a better knowledge of J2EE? > ... in ... ch.qos.logback.core.db.JNDIConnectionSource: > Maybe some sort of "if jndi then don't use user/pass" testing could be done > for better error reporting I guess if the answer to prev. question is positive this could be a worthy idea [2] > Added to that, the JNDI name used in the "jndiLocation" must be > prepended by "java:comp/env", as this is the standard context for java > objects and the JNDIConnectionSource does not prepend the searched values with it. Looks like a nice idea to me. Will not be backwards compatible but logback is not version 1 yet. I suggest that you create a Jira item for this /and maybe attach a patch/ [3] > Another thing I got problems with was slf4j's jcl-over-slf4j-1.5.2.jar. > logback ... initialized too early when using jcl-over-slf4j, before tomcat > makes the JNDI resources available to the web application > ch.qos.logback.core.db.JNDIConnectionSource@67c1a6 - Error while getting > data source javax.naming.NameNotFoundException: > javax.naming.NameNotFoundException: 'comp' can't be found in this context > at org.apache.naming.NamingContext.lookup(NamingContext.java:770) ... > org.apache.commons.logging.impl.SLF4JLogFactory.getInstance(SLF4JLogFactory.java:155) > org.apache.commons.logging.LogFactory.getLog(LogFactory.java:704) > org.apache.catalina.core.ContainerBase.getLogger(ContainerBase.java:381) ... > org.apache.catalina.startup.HostConfig.start(HostConfig.java:1149) Looks pretty weird to me. Diego what folder is jcl-over-slf4j-1.5.2.jar in? Do you have another copy in shared tomcat libraries? Or together with tomcat own jars? Cheers, Anton Tagunov logback user

4 4

Re: [logback-user] Special Joran substitution
by Diego Louzán 15 Jul '08

15 Jul '08

Ceki Gulcu escribió: > Hello Diego, > > The special syntax originates from Bash, at least that's where I saw it fist. > > http://www.faqs.org/docs/abs/HTML/parameter-substitution.html > http://tldp.org/LDP/abs/html/refcards.html#AEN20779 > http://www.purinchu.net/wp/2005/06/11/advanced-parameter-substitution-with-… > > Thus, "${varnname:-word}" means, if varname exists and isn't null, return its > value, otherwise return 'word'. > > If my memory serves me correctly, the special substitution syntax is needed > because, we would like to have the name of the parameter to vary with the > circumstances. More precisely, while most database drivers take expect the name > of the "url" parameter to be "url", some database drivers expect that the value > be passed using a different variable name. > > Does this help? > > Diego Louzán wrote: >> In chapter4 of the documentation the following example is shown: >> >> <dataSource class="${dataSourceClass}"> >>  >> <param name="${url-key:-url}" value="${url_value}"/> >> <serverName>${serverName}</serverName> >> <databaseName>${databaseName}</databaseName> >> </dataSource> >> >> I can't figure out that special syntax for Joran in "${url-key:-url}", >> can anybody explain it to me? Why is it needed in that example? >> >> Regards. >> Diego. I'm not much of a bash fan, so no wonder I had never heard about that syntax xD Thanks for clearing that up. Regards. Diego.

1 0

Problem with FileNamePattern that includes ${user.home} on windows
by Michel Collette 14 Jul '08

14 Jul '08

Hello, I'm been unsuccessful at using the RollingFileAppender with a FixedWindowRollingPolicy that has this FileNamePattern: <FileNamePattern>${user.home}/Client%i.log</FileNamePattern> "${user.home}/Client%i.log" gets correctly expanded to "C:\\Documents and Settings\\Michel/Client%i.log" when it's passed to the FixedWindowRollingPolicy constructor. The problem starts once the filename gets parsed by ch.qos.logback.core.pattern.parser.Parser: the filename becomes "C:ocuments and Settingsichel/Client". File rotations then fails (as would be expected) when trying to rename the current log to "C:ocuments and Settingsichel/Client1.log" Is there a way to configure logback to correctly deal with the "\\" used in windows? Thanks.

2 1

Re: [logback-user] Test logging in unittest
by Ceki Gulcu 14 Jul '08

14 Jul '08

Claus, Thank you for posting your recipe for unit testing. If you wanted, you could inspect the contents of the log events to see if it meets your expectations. For example you could write. assertEquals("Exactly 1 failure should have been logged.", 1, listAppender.list.size()); // inspect message contents: LoggingEvent le = listAppender.list.get(0); assertTrue(le.getFormattedMessage().startsWith("whatever you expect it to start with")); HTH, clanie wrote: > Yes, that's certainly clearer, thank you. > > Just to in case others with the same problem stumbles upon this thread, > here's the new improved version the test: > > /** > * Tests that closeChannels() tries to close all the supplied Channels, > * even if supplied with nulls and Channels which fail to close. > */ > @Test > public void testCloseChannels() { > // Capture log entries from FileUtil > Logger logger = (Logger) LoggerFactory.getLogger(FileUtil.class); > logger.setAdditive(false); > ListAppender<LoggingEvent> listAppender = > new ListAppender<LoggingEvent>(); > listAppender.start(); > logger.addAppender(listAppender); > // Close a mix of normal, broken and completely missing channels > closeChannelse(null, new UnclosableChannelStub(), > null, new ChannelStub(), new ChannelStub()); > // Restore normal logging > logger.detachAppender(listAppender); > logger.setAdditive(true); > // Perform checks > assertEquals("Exactly 1 failure should have been logged.", 1, > listAppender.list.size()); > } > > > > > Ceki Gulcu-2 wrote: >> >> Hello Claus, >> >> A slightly easier way would be to use a ListAppender, which appends each >> event >> into a ...list. Logback-core has an implementation of about 10 lines each >> which >> is extensively for testing purposes in the other modules. >> >> Here it is: >> >> package ch.qos.logback.core.read; >> >> import java.util.ArrayList; >> import java.util.List; >> >> import ch.qos.logback.core.AppenderBase; >> >> public class ListAppender<E> extends AppenderBase<E> { >> >> public List<E> list = new ArrayList<E>(); >> >> protected void append(E e) { >> list.add(e); >> } >> } >> >> Check the tests where ListAppender is used. I think BasicLoggerTest should >> give >> you a good idea of usage. >> >> >> >> HTH, >> >> Claus Nielsen wrote: >>> I have a unit test which checks that a class writes the log statements >>> it's supposed to. >>> >>> The following code works, but I was wondering if there isn't a better or >>> easier way to do this. >>> >>> Here's what I have now: >>> >>> public void testCloseChannels() { >>> Logger rootLogger = (Logger) >>> LoggerFactory.getLogger( >>> LoggerContext.ROOT_NAME); >>> Appender<LoggingEvent> appender = >>> rootLogger.getAppender("console"); >>> appender.addFilter(new Filter() { >>> @Override >>> public FilterReply decide(Object o) { >>> LoggingEvent event = (LoggingEvent) o; >>> if (event.getMessage().equals("Failed to close channel.")) { >>> expectedLogEntryCount.incrementAndGet(); >>> return FilterReply.DENY; >>> } >>> return FilterReply.NEUTRAL; >>> } >>> }); >>> close(null, new UnclosableChannelStub(), null, >>> new ChannelStub(), new ChannelStub()); >>> assertEquals("Exactly 1 failure should have been logged.", >>> 1, expectedLogEntryCount.intValue()); >>> appender.clearAllFilters(); >>> } >>> >>> >>> Regards, >>> Claus Nielsen >> -- >> -- >> Ceki Gülcü >> QOS.ch is looking to hire talented developers located in Switzerland. >> If interested, please contact c e k i AT q o s . c h >> _______________________________________________ >> Logback-user mailing list >> Logback-user(a)qos.ch >> http://qos.ch/mailman/listinfo/logback-user >> >> > -- -- Ceki Gülcü QOS.ch is looking to hire talented developers located in Switzerland. If interested, please contact c e k i AT q o s . c h

1 0

Enabling debug for one Marker only ?
by Franck Routier 26 Jun '08

26 Jun '08

Hi, I have seen there is a logger.isDebugEnabled(Marker marker) method. How would I configure I want debug enbled for one Marker only ? Something like : <logger name="mylogger" marker="mymarker"> <level value="DEBUG" /> </logger> I could find the releant documentation. Or should I play with filters ? (but as I understood it, filters happen after the log statement has been issued). Franck

1 0

Logback -> Chainsaw
by Lilianne E. Blaze 06 Jun '08

06 Jun '08

Hello, Is there any official / recommended way to connect (via TCP) Logback to Log4j's Chainsaw? Greetings, Lilianne E. Blaze

1 0

Re: [logback-user] logback equivalent of %c{1} in log4j?
by Arthur Blake 05 Jun '08

05 Jun '08

Sorry about the messed up JIRA summary line... It was a cut & paste error and I couldn't figure out how to edit the summary after the issue was already created... It should have been simply: "PatternLayout needs to be able to show class name of the logger without it's package prefix" ----- Original Message ---- From: Ceki Gulcu <listid(a)qos.ch> To: logback users list <logback-user(a)qos.ch> Sent: Thursday, June 5, 2008 10:19:31 AM Subject: Re: [logback-user] logback equivalent of %c{1} in log4j? Thank you. Arthur Blake wrote: > No problem: > > http://jira.qos.ch/browse/LBGENERAL-19 > > ----- Original Message ---- > From: Ceki Gulcu <listid(a)qos.ch> > To: logback users list <logback-user(a)qos.ch> > Sent: Thursday, June 5, 2008 8:44:22 AM > Subject: Re: [logback-user] logback equivalent of %c{1} in log4j? > > > Hello Arthur, Silvano, > > Arthur, could you please enter a Jira issue referencing this thread as > well as > the thread from last year so that this feature can be added in the near > future. > > Is that OK with you? > > Arthur Blake wrote: > > Unfortunately there isn't... > > I asked the same question almost a year ago: > > > > See http://marc.info/?l=logback-user&m=118426697600168&w=2 > <http://marc.info/?l=logback-user&m=118426697600168&w=2> > > _______________________________________________ Logback-user mailing list Logback-user(a)qos.ch http://qos.ch/mailman/listinfo/logback-user

1 0

Re: [logback-user] logback equivalent of %c{1} in log4j?
by Arthur Blake 05 Jun '08

05 Jun '08

No problem: http://jira.qos.ch/browse/LBGENERAL-19 ----- Original Message ---- From: Ceki Gulcu <listid(a)qos.ch> To: logback users list <logback-user(a)qos.ch> Sent: Thursday, June 5, 2008 8:44:22 AM Subject: Re: [logback-user] logback equivalent of %c{1} in log4j? Hello Arthur, Silvano, Arthur, could you please enter a Jira issue referencing this thread as well as the thread from last year so that this feature can be added in the near future. Is that OK with you? Arthur Blake wrote: > Unfortunately there isn't... > I asked the same question almost a year ago: > > See http://marc.info/?l=logback-user&m=118426697600168&w=2 > > ----- Original Message ---- > From: Silvano Maffeis <silvano(a)maffeis.com> > To: logback-user(a)qos.ch > Sent: Thursday, June 5, 2008 7:42:25 AM > Subject: [logback-user] logback equivalent of %c{1} in log4j? > > Hi > > In my logs I would like to see only the name of the class issuing the > log statement, > without any package information. In log4j I use %c{1} for that. In > logback I tried > various patterns, such as %logger{30}, but the best I can get is > the abbreviated package name in front of the class name, e.g., > something like > "c.v.t.s.session.SessionPlugin". > > Is there a pattern for getting only the class name without any package > information ? > > Cheers, > Silvano _______________________________________________ Logback-user mailing list Logback-user(a)qos.ch http://qos.ch/mailman/listinfo/logback-user

2 1

Re: [logback-user] logback equivalent of %c{1} in log4j?
by Arthur Blake 05 Jun '08

05 Jun '08

Unfortunately there isn't... I asked the same question almost a year ago: See http://marc.info/?l=logback-user&m=118426697600168&w=2 ----- Original Message ---- From: Silvano Maffeis <silvano(a)maffeis.com> To: logback-user(a)qos.ch Sent: Thursday, June 5, 2008 7:42:25 AM Subject: [logback-user] logback equivalent of %c{1} in log4j? Hi In my logs I would like to see only the name of the class issuing the log statement, without any package information. In log4j I use %c{1} for that. In logback I tried various patterns, such as %logger{30}, but the best I can get is the abbreviated package name in front of the class name, e.g., something like "c.v.t.s.session.SessionPlugin". Is there a pattern for getting only the class name without any package information ? Cheers, Silvano _______________________________________________ Logback-user mailing list Logback-user(a)qos.ch http://qos.ch/mailman/listinfo/logback-user

2 1

logback equivalent of %c{1} in log4j?
by Silvano Maffeis 05 Jun '08

05 Jun '08

Hi In my logs I would like to see only the name of the class issuing the log statement, without any package information. In log4j I use %c{1} for that. In logback I tried various patterns, such as %logger{30}, but the best I can get is the abbreviated package name in front of the class name, e.g., something like "c.v.t.s.session.SessionPlugin". Is there a pattern for getting only the class name without any package information ? Cheers, Silvano

1 0