Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Bigtop >> mail # user >> Customizing pom behaviour across different modules

Copy link to this message
Re: Customizing pom behaviour across different modules
On Fri, Aug 23, 2013 at 4:48 AM, Jay Vyas <[EMAIL PROTECTED]> wrote:
> Hi bigtop:
> I have hacked a bigtop test cycle together which builds and tests from my source.

Being able to run from source somehow would be a *really* good thing to have.
After all, most of our tests are implemented in Groovy, but even for Java we
should be able to have a script that just 'does the right thing' given
the source
repo and runs the tests.

Not being able to do that is the biggest challenge we have compared to the
testsuites implemented in bash/python/etc.

> It is a bit of a hack because:
> 1) It consists of 2 build phases first is online, to pull deps and install snapshots
> and second one offline, to overwrite snapshots with local source.
> 2) it modifies the <modules> tags in test-artifacts and smokes so that the build only
> builds a couple of submodules and only tests a couple of submodules.
> Id like the script to be cleaner, maybe by sending modules in at the commandline and
> having a simpler way to deal with snapshots.
> Any hints on this?

I'd suggest filing a JIRA and attaching your work in progress patches there so
that folks can comment and share feedback.

> Any hints on ways to customize the pom/build without editing and
> maintains changes to the core pom xml files? Of maybe example scripts used for
> running bigtop smokes in production?

It is a bit difficult to give a focused advise without actually looking
at the code that you've got already. As I said -- please file a JIRA.