Github user aledsage commented on a diff in the pull request:
https://github.com/apache/brooklyn-server/pull/734#discussion_r124584831
--- Diff:
core/src/main/java/org/apache/brooklyn/core/location/AbstractLocation.java ---
@@ -215,6 +216,16 @@ public AbstractLocation configure(Map<?,?> properties)
{
config().removeKey(PARENT_LOCATION);
}
+ // allow config keys to be set by name (or deprecated name)
+ //
+ // Aled thinks it would be sensible to remove the consumed flags
below (i.e. properties = ...).
+ // However, that caused ClockerDynamicLocationPatternTest to fail
because there is a field of
+ // StubContainerLocation annotated with `@SetFromFlag("owner")`,
as well as a config key with
--- End diff --
Not certain - I deleted the test that used it in #728. My concern is
whether that test highlighted that we're supposed to be supporting such madness!
@ahgittin's comment was: "if it's not too hard to log a warning wherever we
rely on it, however, that would be ideal. if that's not easy then fine to
abandon.".
It's probably tricky to log.warn on it reliably (because
`FlagUtils.setFieldsFromFlagsWithBag` and `FlagUtils.setAllConfigKeys` also
does logic for setting config, so we can't easily tell how it's using the key).
We could try adding some new code, e.g. in `EntityDynamicType`, to detect
it, and warn (or even fail) if we see the name being used in multiple places. I
suggest we do that separately from this PR, and I suggest that we fail-fast.
I'd be happy with abandoning - i.e. change it to what I proposed in the
comment, to not pass the used properties to
`FlagUtils.setFieldsFromFlagsWithBag` or `FlagUtils.setAllConfigKeys`.
I'll go with abandoning, unless anyone feels strongly to the contrary?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
