Re: Want something like "failonerror", for artifactspublisher

Hmmm... so Publish failure will never cause Build failure, eh? So much for 
that approach. Was hoping for a "catch-all" at Publish time, but I guess 
that's not going to happen.

I'll have to try to find a way to cause build failure, as you say. And, yes, 
build is with Ant.

Thanks for the investigation.

On 8/13/2010 10:38 AM, Dan Rollo wrote:
> Hi Doug,
>
> I may be missing the point here, but looking at the source code of the
> Project.publish() method, no publisher failure should cause the build to
> be treated as "failed". By the time the "publish" phase begins, by
> definition, the "build" phase has completed (and has either failed or not).
>
> http://cruisecontrol.svn.sourceforge.net/viewvc/cruisecontrol/trunk/cruisecontrol/main/src/net/sourceforge/cruisecontrol/Project.java?revision=4521&view=markup
>
>
>       /**
>        * Iterate over all of the registered<code>Publisher</code>s and call
>        * their respective<code>publish</code>  methods.
>        * @param buildLog the content to publish
>        * @throws CruiseControlException if an error occurs during publishing
>        */
>       protected void publish(final Log buildLog) throws
> CruiseControlException {
>           setState(ProjectState.PUBLISHING);
>           for (final Publisher publisher : projectConfig.getPublishers()) {
>               // catch all errors, Publishers shouldn't cause failures in
> the build method
>               try {
>                   publisher.publish(buildLog.getContent());
>               } catch (Throwable t) {
>                   final StringBuilder message = new
> StringBuilder("exception publishing results");
>                   message.append(" with
> ").append(publisher.getClass().getName());
>                   message.append(" for project ").append(name);
>                   LOG.error(message.toString(), t);
>               }
>           }
>       }
>
> One thought - if you know why the publish is failing (eg: a generated
> file is missing), have you tried adding a check for that file to the
> "build" phase for the project?
> If using an Ant builder, you could do this with a small change to the
> build.xml to check for the required file to publish.
> Another approach is to use a Composite builder, where the last builder's
> job is to check for the needed file.
>
> The idea is to detect the error condition during the "build" phase,
> where you can still affect the "pass/fail" status of the overall build.
>
> Dan
>
>
> Date: Wed, 11 Aug 2010 17:56:17 -0500
> Subject: Re: [Cruisecontrol-user] Want something like "failonerror"
> 	for	artifactspublisher
> To: [email protected]
>
> Bah, my Ant copy still didn't fail when the file being copied didn't exist.
>
> 	<copy todir="${publishDest}/${projectname}/${cctimestamp}">
> 		<fileset dir="${publishSource}">
> 			<include name="${publishInclude}"/>
> 		</fileset>
> 	</copy>
>
>
> On 7/26/2010 2:02 PM, Marc W. wrote:
>   >  >  I found myself replacing my antpublisher tasks with an Ant script
> called by
>   >  >  antpublisher.  I simply used the passed property, cctimestamp, to
> locate the
>   >  >  proper artifact directory within Ant.
>   >  >
>   >  >  I'm currently working on getting Ant's output logged (or copied) to the
>   >  >  appropriate build's artifacts directory though.
>   >  >
>   >  >  HTH,
>   >  >  -Marc
>   >  >
>   >  >
>   >  >  On Mon, Jul 26, 2010 at 11:50, Doug Breaux
> <[email protected]
>   >  >  <mailto:[email protected]>>  wrote:
>   >  >
>   >  >      So any advice/ideas here? I still need to see a CC build
> failure if the
>   >  >      artifactspublisher can't find its file(s) to copy.
>   >  >
>   >  >      On 5/26/2010 8:33 AM, [email protected]
>   >  >      <mailto:[email protected]>  wrote:
>   >  >       >  I just finished dealing with a similar problem with the CVS
>   >  >      bootstrapper. The
>   >  >       >  CVS update generated some warnings, which caused the process
> to return a 1
>   >  >       >  instead of 0. This prevented the subsequent ant build from
> running. Since
>   >  >       >  there is no notification tied to the CVS failure, this
> problem went
>   >  >      unnoticed
>   >  >       >  for over a week.
>   >  >       >
>   >  >       >  Joel
>   >  >       >
>   >  >       >  -----Original Message-----
>   >  >       >  From: "Assaf Lavie"<[email protected]
> <mailto:[email protected]>>
>   >  >       >  Sent: Wednesday, May 26, 2010 2:44am
>   >  >       >  To: [email protected]
>   >  >      <mailto:[email protected]>
>   >  >       >  Subject: Re: [Cruisecontrol-user] Want something like
> "failonerror" for
>   >  >       >  artifactspublisher
>   >  >       >
>   >  >       >  What sort of build task isn't failing on error? I'm aware of
> such a problem
>   >  >       >  with Ant, for example, but it's fixable.
>   >  >       >
>   >  >       >  On Mon, May 24, 2010 at 8:38 PM, Doug Breaux
>   >  >      <[email protected]<mailto:[email protected]>
>   >  >       >  <mailto:[email protected]
>   >  >      <mailto:[email protected]>>>  wrote:
>   >  >       >
>   >  >       >      Is there any way to get CC to consider the build failed
> if the
>   >  >       >      artifactspublisher fails? It's logging an error about
> not being
>   >  >      able to copy
>   >  >       >      the file in question, but CC still says the build was
> successful.
>   >  >       >
>   >  >       >      (I have some build-type tasks that are not causing the
> build to
>   >  >      fail when they
>   >  >       >      fail, despite failOnError="true", and it seems one of
> the simplest
>   >  >      ways to
>   >  >       >      catch all of them is to report a failure when the
> artifact doesn't
>   >  >      exist.)
>   >  >       >
>   >  >       >      Doug

------------------------------------------------------------------------------
This SF.net email is sponsored by 

Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev