The commons wrapper we were using didn't support executable files
so Ant seems like the best choice ultimately, even if it has a
lot of features we don't use or need.
There is an issue with Kafka in the M1 release (the app generated
here would blow up because of a mismatch between the client and
server). Snapshots are best for now, until we hit RC1.
If we don't re-use an existing factory, the label doesn't change and
all the snippets get generated with the same identifier.
Remove explicit version from wiremock test
Add snippets attribute to build config
See gh-295
This commit commit adds restdocs and stub generators and initiate
a reference guide for Initializr.
Most of the controller tests now use MockMvc via a custom version
of the MockMvcClientHttpRequestFactory (from spring-test). The
snippet names are auto-generated in the form
<HttpMethod>/<path>[/queries(/<name-value)*][/headers](/name-value)*]
when there is a comma-separated value in a header it is
abbreviated as <first-value>.MORE.
Wiremock stubs are generated in the same form under
snippets/stubs (with ".json" as the
file extension).
The controller tests that stayed as full stack use a different
base class AbstractFullStackInitializrIntegrationTests.
A long JSON body can be broken out into separate snippets
for each field (or rather a list of fields supplied by the
user). This feature was already used with hard-coded snippets
in the wiki.
See gh-295
The additional bom fixes the version of spring cloud task
(which is different than the one in the main cloud BOM).
Fixes gh-342 some more.
Closes gh-344
If a RedisConnectionFactory is detected the system will now export
metrics to redis every 5s (initializr.metrics.rateMillis). Metric
names are prefixed with <id>.<hex>. where <id> defaults to the
application name (initializr.metrics.id) and <hex> is unique. An
aggregator app can then pick up the values and (for instance)
continue counters across restarts.
The code is meant to be easier to upgrade Spring Boot 1.3
There are 2 beans that we can simply remove when we upgrade to 1.3.
Meanwhile, the app will run with 1.3 (or 1.2) for testing if you
uncomment the @ExportMetricWriter.
The instructions should be explicit and precise enough for
anyone to deploy to production now. There is a Bamboo job
that should be doing it, but we've had some issue with it and
it often fails, whereas manual pushes always work. You need
a Cloudfoundry account with access to the Sagan organization
to push to production (ask Trevor Marshall).
Dave Syer