Brett Porter wrote:
On 6/8/05, Steve Loughran <[EMAIL PROTECTED]> wrote:
Here in ant-land, we've done two things to make the stuff work better
1.
Are these the same points you added below?
no, I forgot to type some stuff
1. Resource collections are a new way of representing stuff; filesets
are inherently resource collections. I'm told this will make it easier
to manipulate filesets.
2. what was #2? I forget.
1. build file tests; check handling of missing required items, etc, etc.
ant-testutils is the jar for this.
Ok, that will be worth a look. I'm pretty flat out for the next ocuple
of weeks, then JavaOne, then holidays - so I'll revisit this next
month :)
Is this still going to give enough time to prepare for Ant 1.7?
Maybe. I might have a go at starting the test suite, its not too hard in
junit, and now AntUnit promises more simplicity.
5. something to map from (artifact,groupid,version) to the path to a
jar, to a pom, or the base dir. Here is my macro to get the dest
directory, used when copying in local stuff.
I thought that was why dependencies had a fileset ID, so you didn't
need to worry about constructing paths, or knowing where the local
repository is if using external config?
its primarily for local deployment-by-copying. Now, maven itself has a
built in publish-to-a-repository goal doesnt it? That is what i really
need, I suspect. Push out an artifact to a location, generate a POM file
on demand, plus the checksums.
Anyway, happy for this to go into JIRA. It can use the code already in
Maven to set the property and it would be flexible across layouts.
The proxy idea in a later email is also good, JIRA'd.
...just another example of how J2SE is laptop unfriendly. I think it
comes from sun's no-laptop-except-powerbooks-you-buy-yourself policy.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
