You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@maven.apache.org by "Brian E. Fox" <br...@reply.infinity.nu> on 2008/12/30 20:18:37 UTC
Javadoc is back to it's old misbehaving (forking) ways
We had lots of problems with javadoc 2.3 that introduced forking behavior.
This was reverted in 2.4 and apparently put back in 2.5:
http://svn.apache.org/viewvc/maven/plugins/tags/maven-javadoc-plugin-2.5/sr
c/main/java/org/apache/maven/plugin/javadoc/JavadocReport.java?r1=661429&r2=
661899
This causes problems in some projects that are now forking for example
during a release. Can we find a better way to fix this without making it
fork?
(btw this is exactly one of the issues we need to fix in maven 3, a plugin
needs to declare that at least a given phase should be run and thus if bound
to a phase it won¹t fork, but if run from the cli directly, it will)
http://jira.codehaus.org/browse/MRELEASE-271
http://jira.codehaus.org/browse/MJAVADOC-104
Re: Javadoc is back to it's old misbehaving (forking) ways
Posted by Benjamin Bentmann <be...@udo.edu>.
Brian E. Fox wrote:
> The trouble is the javadoc goal is bound in the release profile which is getting activated when people do releases. IMO we should put javadoc back the way it was in 2.4 [...]
I don't think that r661899 is the cause for the regression and hence it
wouldn't make sense to revert it.
AFAIK, the problem is not forking by itself but the combination of
forking with aggregating. For instance, the source:jar goal employed by
the release profile forks, too, but to my knowledge doesn't cause
problems. In Javadoc Plugin version 2.5, neither javadoc:jar (invoked
during "package" phase) nor javadoc:javadoc (invoked during "site"
phase") are forking aggregators. The only forking aggregators are the
new javadoc:*aggregate goals.
Unless we have a build log or an IT to prove that javadoc:*javadoc is
causing the regression, I feel we're better off to tweak the new goals.
Benjamin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
RE: Javadoc is back to it's old misbehaving (forking) ways
Posted by "Brian E. Fox" <br...@reply.infinity.nu>.
I know we vacillated back and forth on these issues. The trouble is the javadoc goal is bound in the release profile which is getting activated when people do releases. IMO we should put javadoc back the way it was in 2.4 and make the new goal as intended to be used from the cli. This way it just works for most people doing releases, and for those that actually use source generation, they use the other goal. I don't think the special new goal should have the functionality most people actually need.
-----Original Message-----
From: Vincent Siveton [mailto:vincent.siveton@gmail.com]
Sent: Monday, January 05, 2009 3:17 PM
To: Maven Developers List
Subject: Re: Javadoc is back to it's old misbehaving (forking) ways
Hi Brian,
In fact, MJAVADOC-168 readds the original behaviour for generated sources.
MJAVADOC-104 was an @aggregator pb and this was reverted in 2.4. In
2.5, we have javadoc:aggregate so try with this goal.
Cheers,
Vincent
2008/12/30, Brian E. Fox <br...@reply.infinity.nu>:
> We had lots of problems with javadoc 2.3 that introduced forking behavior.
> This was reverted in 2.4 and apparently put back in 2.5:
> http://svn.apache.org/viewvc/maven/plugins/tags/maven-javadoc-plugin-2.5/sr
> c/main/java/org/apache/maven/plugin/javadoc/JavadocReport.java?r1=661429&r2=
> 661899
>
> This causes problems in some projects that are now forking for example
> during a release. Can we find a better way to fix this without making it
> fork?
>
> (btw this is exactly one of the issues we need to fix in maven 3, a plugin
> needs to declare that at least a given phase should be run and thus if bound
> to a phase it won¹t fork, but if run from the cli directly, it will)
>
> http://jira.codehaus.org/browse/MRELEASE-271
> http://jira.codehaus.org/browse/MJAVADOC-104
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
Re: Javadoc is back to it's old misbehaving (forking) ways
Posted by Vincent Siveton <vi...@gmail.com>.
Hi Brian,
In fact, MJAVADOC-168 readds the original behaviour for generated sources.
MJAVADOC-104 was an @aggregator pb and this was reverted in 2.4. In
2.5, we have javadoc:aggregate so try with this goal.
Cheers,
Vincent
2008/12/30, Brian E. Fox <br...@reply.infinity.nu>:
> We had lots of problems with javadoc 2.3 that introduced forking behavior.
> This was reverted in 2.4 and apparently put back in 2.5:
> http://svn.apache.org/viewvc/maven/plugins/tags/maven-javadoc-plugin-2.5/sr
> c/main/java/org/apache/maven/plugin/javadoc/JavadocReport.java?r1=661429&r2=
> 661899
>
> This causes problems in some projects that are now forking for example
> during a release. Can we find a better way to fix this without making it
> fork?
>
> (btw this is exactly one of the issues we need to fix in maven 3, a plugin
> needs to declare that at least a given phase should be run and thus if bound
> to a phase it won¹t fork, but if run from the cli directly, it will)
>
> http://jira.codehaus.org/browse/MRELEASE-271
> http://jira.codehaus.org/browse/MJAVADOC-104
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org