Hi Rob,

at the moment I'm busy upgrading the maven-release-plugin to Maven3 (i.e. dropping Maven2 support) That meant rewriting quite some unittests and upgrading the maven-plugin-testing-harness (current one is really old).

In some cases the maven-plugin-testing-harness is hard to use, so most of the time we're using the maven-invoker-plugin to make the integration tests. These tests are a little bit harder to debug from IDE when required, but they should fit better with the way the plugin is executed by Maven.

thanks,
Robert

On Sat, 06 Jan 2018 20:03:20 +0100, Rob Tompkins <[email protected]> wrote:

Hello again Maven Team,

I’ve made substantive progress towards getting the commons team a release plugin (https://github.com/chtompki/commons-release-plugin <https://github.com/chtompki/commons-release-plugin>). However, in my attempt at writing plugin unit tests for the first time, I’ve found myself running into difficulties in dealing with which dependencies need to be in the classpath to effectively run the maven-plugin-testing-harness.

I was hoping to get another set of eyes on my work, namely the pom and the unit test that is in flight (see the above repository), such that I can get around these classpath issues and start writing proper tests for the plugin. It is quite easy to see the issues that I’m having by cloning the project and running “mvn test:"

Running org.apache.commons.release.plugin.mojos.CommonsSiteCompressionMojoTest
Jan 06, 2018 2:00:13 PM org.eclipse.sisu.inject.Logs$JULSink warn
WARNING: Error injecting: org.apache.maven.artifact.resolver.DefaultArtifactResolver java.lang.NoClassDefFoundError: Lorg/apache/maven/artifact/transform/ArtifactTransformationManager;
        at java.lang.Class.getDeclaredFields0(Native Method)
        at java.lang.Class.privateGetDeclaredFields(Class.java:2583)
        at java.lang.Class.getDeclaredFields(Class.java:1916)

Any guidance here would be much welcomed, and moreover I can’t thank you guys enough for the previous insights because they gave me the ability to make solid progress on our release automation.

Many thanks and all the best,
-Rob


On Dec 28, 2017, at 4:53 PM, Rob Tompkins <[email protected]> wrote:



On Dec 28, 2017, at 4:05 AM, Hervé BOUTEMY <[email protected]> wrote:

Hi Rob,

one additional point: currently, for Maven itself, instead of adding new Maven-specific ReleasePhase(s) to the default configuration, or configure them in our parent pom (I'm not sure documentation on how to do that is available: I could not find it), we chose first to create a separate "dist-tool" to check consistency of what is currently published in misc places and provide some
commands to fix when an inconsistency is found.

This happens through daily reports done by a Jenkins job [1]:
- distribution area vs Maven Central [2]
- version from Maven site vs Maven Central [3]

We did not produce any release nor made it really configurable for conventions different from Maven ones (like Common's -src & -bin), but at least there is a
configuration file to define artifacts to check [4]

Interesting. Thanks,.

-Rob


HTH

Hervé


[1] https://builds.apache.org/view/M-R/view/Maven/job/dist-tool-plugin/

[2] https://builds.apache.org/view/M-R/view/Maven/job/dist-tool-plugin/site/
dist-tool-check-source-release.html

[3] https://builds.apache.org/view/M-R/view/Maven/job/dist-tool-plugin/site/
dist-tool-check-index-page.html

[4] https://builds.apache.org/view/M-R/view/Maven/job/dist-tool-plugin/site/
dist-tool.conf.html

Le mercredi 27 décembre 2017, 23:32:49 CET Rob Tompkins a écrit :
Stephen,

I can’t thank you enough for your reply. I’ll take your suggestions and continue to sandbox around using the maven-release-plugin as a guideline.

All the best and happy holidays,
-Rob

On Dec 26, 2017, at 5:27 AM, Stephen Connolly
<[email protected]> wrote:>
On Tue 26 Dec 2017 at 03:10, Rob Tompkins <[email protected]
<mailto:[email protected]>> wrote:
Hello all,

Pardon, maybe this should have gone to your @user list, but why not ping the dev crew. I’ve been playing around the ideas surrounding our fairly manual release process for the components in Commons, and I was hoping
for
some insights.

Scripting the version changes isn’t really that big of a deal for us, and that I can manage. But, when it comes to publishing our artifacts to the apache nexus repository, and then separately publishing our -src and -bin
assemblies to the dev dist subversion repository (and consequently
deleting
those artifacts from nexus as they’re “attached” for the purpose of gpg
signing), I feel it a tad cumbersome.

I’ve fiddled around a little with the idea of detaching the -src and -bin assemblies after gpg signing with some success, but then I have to delve into the mechanics of publishing those up to the subversion repository,
and
clearly that problem has already been solved.

Is your problem you don’t want those going to Nexus staging but only to
dist? Or is it that you want them *also* going to dist.

Personally... I see no reason to remove them from going to Nexus staging
(in fact I have a background plan to add secondary signing support to
staging... i’m Waiting to see the Nexus 3 staging APIs before attempting
though. That would mean that the PMC would be able to *add* their GPG
signature to the staged artifacts as part of the voting... in which case you’d want to hold off uploading to dist until *after* the vote so you get
the full set of signatures)

If you want to upload a subset of attached artifacts to dist as part of
the
release, that seems much more tenable... just a derivative of the
scm-publish plugin from what I can see.

So I find myself in the space of trying to shoehorn our process into its

the main maven-release-plugin, which I’ve found a tad difficult, versus writing our own release plugin, which feels like I would be duplicating
tons of code (which I don’t want to do).

So the release plugin is really two parts:

1. A toolkit for writing release plugins

2. An example that does the job for the requirements of the Maven TLP and
has seemed “sufficient” for a lot of other people.

As such, if you have different needs, do not feel bad about having to
encode differently... hopefully the toolkit half of the codebase is
sufficient for you.

I’m curious if you guys have any thoughts on the matter as I’ve been
playing around in the space for a little while now.

Cheers and happy holidays from UTC-5,
-Rob
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
<mailto:[email protected]> For additional commands,
e-mail: [email protected] <mailto:[email protected]>

--

Sent from my phone

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to