Differences Between Packaging and Dependency Type
Ref: http://jira.codehaus.org/browse/MNG-257.
I'd like to work through any problems that might arise from the
differences between <packaging> and <type>. Below is the background,
problem statement, and design decision to rectify them. Please add any
additional problems you see, or issues with the design.
Background
In Maven 2.0, projects are keyed uniquely on only groupId and artifactId
for considering what is a project (version does come into it for
determining different versions of a project).
A packaging is specified on pom.xml to signify how to build that
project, and what type of artifact it produces.
For a given packaging, it may produce multiple artifacts that share the
same project information - the main example used is an ejb that also
outputs an ejb-client. There are also some goals that will produce
artifacts from a given project - for example, the assembly/distribution
goals, javadoc and sources.
To be able to depend on only some of those artifacts, <type> on a
dependency may not match packaging - if you declare <type>ejb</type> you
only get the ejb, and if you declare <type>ejb-client</type>, you only
get the ejb-client JAR. The POM can always be located by the
groupId/artifactId/version combination (type is not present in the m2
repo, and the POM is always under /poms/ if using the old layout).
This also means that while type is not considered in whether an project
is unique, it is for a dependency - so you can download tld files with
the same name as the JAR they belong to.
This is pretty sound from general use cases, though there have been a
couple of problems to arise:
- inconsistency could lead to confusion
- guaranteeing snapshots match versions
Inconsistency:
I believe this is ok. The naming was different to ensure it is clear
that they can be different, and packaging IS A SUBSET of <type>,
guaranteeing they can be handled consistently. Any value of packaging
should have 1 type that matches exactly, and possibly more types that
are associated with it. Any type has 0..1 matching packaging.
Snapshots:
If a JAR is deployed, then a distribution is deployed - either the POM is
republished (so the snapshot is bumped up and the JAR is orphaned), or
it is not and the published dist is orphaned, or reuses the timestamp
which is not technically correct (especially as it may overwrite a
previous distribution).
Discussion
Firstly, we differentiate types in the following way:
- that specified by packaging
- other types "tied" to the packaging (eg ejb-client and ejb)
- other types only generated by particular optional goals (eg distribution)
Untied objects can be manually attached, and tied objects can be manually detached. From this, it seems sensible that attachment is only a matter of configuration, with sensible defaults, rather than something specific in the design. Note that there are different scenarios: attachment by configuration is appropriate for ejb client, but when set it should always happen. For distributions, sources, etc it is appropriate to only attach at release - they do not need to be deployed for snapshots.
- deploying a distribution MUST deploy the JAR - ie, any deployment must
deploy both the POM and the artifact of the packaging (+ anything tied
to it, like an ejb-client or sources). - deploying an artifact MUST deploy any "tied" artifacts - eg, deploying
an EJB must deploy its client - deploying an artifact NEED NOT deploy any "untied" artifacts - eg,
deploying a JAR doesn't deploy the distribution by default - it must be
configured to do so, or the goal added
This ensures that a POM is always deployed and updated for every
deployment of anything, and that the primary artifact and anything tied
to it is never orphaned.
It can mean that under the default lifecycle, there will be no snapshots
deployed for a distribution, for example. This simply means you cannot
depend on a distribution where you can a JAR, unless it is tied, which
is reasonable.
This does not affect released artifacts, as they can be released at
different times and still have the same version.
Declaring the type in a dependency
It is envisaged that the type could encapsulate the extra information about the classifier. ie, ejb-client is a type that maps to the packaging ejb with the classifier "client".
However, there are a few interesting points here:
- have decided to make groupId/artifactId unique, regardless of type. Therefore, type is not required to lookup an artifact at all (though it would be helpful in indicating what type the artifact is when reading the pom)
- the above could mean that type could be replaced by by something solely to identify the classifier
- if not, then either the type needs to map to a packaging + classifier, or a classifier field needs to be added
- if the type is being mapped, this needs to be quite flexible - ie a plugin can provide its own, but this means that arbitrary classifiers can't be used. This has positives and negatives of its own.
I am in favour of using type mappings, at least to start with. The positives are:
- avoids arbitrary classifiers, making artifactId be well considered
- avoids requirement of additional metadata field in the dependency element
Drawbacks: - can't create your own classifiers, must stick to those provided by the plugin producing the artifact
- must build in the mapping to all plugins providing a packaging.
Summary of Solution
We need:
- building an attachment (distribution primarily) must build the original, and deploy both.
- type to packaging + classifier mappings (but no need for the reverse)
- binding for the package step (eg ejb-client, tld) - this will also be applicable to deployment, snapshot or not
- binding for the release step (eg sources, javadoc, distributions, etc)
Decision to make here... Rather than going for the forked lifecycle, is it more appropriate to find a nice way to blend it into the main build?
Building an attachment must build both
TODO
Type -> packaging + classifier mapping
TODO
Attachment at packaging step
Here, a simple call to project.attachArtifact( Artifact artifact ) is appropriate. In the specific use cases:
- ear:ear would create the client based on configuration, and attach it
- a webapp plugin could have a goal, bound to packaging, that when configured published a set of tlds as well as bundled them into a jar
Attachment at release step
We want attachment of sources and javadoc to be the default, but we also want to configure in the attachment of various assemblies.
TODO