Update `ConfigDataEnvironmentContributor.isActive` so that unbound
imports are no longer considered active. Prior to this commit, any
`ConfigDataEnvironmentContributor` that had `null` properties was
considered active. This is incorrect for `Kind.UNBOUND_IMPORT`
contributors since we haven't yet bound the `spring.config.*`
properties.
The `ConfigDataEnvironmentContributorPlaceholdersResolver` has been
updated to handle the refined logic. A placeholder can now be resolved
from the current contributor, or from an unbound contributor by binding
it on the fly.
Fixes gh-29386
Update `SpringBootContextLoader` so that when possible the
`SpringApplication` remains in control of creating the `Environment`
instance.
Prior to this commit, we would always create the `Environment` in the
`SpringBootContextLoader` and then call `setEnvironment` on the
`SpringApplication`. This meant that the `ApplicationEnvironment`
classes were not used and that `isCustomEnvironment` was set to `true`
so no conversion was applied.
With the updated code, an `ApplicationListener` is used to mutate the
`Environment` instance and add the required test property sources.
Fixes gh-29169
We've seen some problems, particularly on CI, where Embedded Mongo
hangs while trying to start the Mongo process. To limit the problem,
this commit replaces the use of Embedded Mongo with Testcontainers
where we can, leaving Embedded Mongo in use only where we're
specifically testing our Embedded Mongo support.
Closes gh-28843
Refine the logic introduced in 64270eca to use a side-effect free
Environment implementation rather than converting the Environment early.
Early conversion can cause condition evaluation issues if
`src/test/resources/application.properties` files are bound to the
`SpringApplication`. Specifically the `spring.main.web-application-type`
property can change the `Environment` type which must happen before
conditions are evaluated.
Fixes gh-29169
Update `launch.script` so that the `run_user` variable is only set when
the user actually exists. Prior to this commit, if the jar file was
owned by a non-existing user the startup script could fail.
Fixes gh-29370
Previously, when writing a repackaged jar, an attempt to write all of
an entry's parent directories would always be made, irrespective of
whether or not the entry itself had already been written. This was
inefficient as, due to the way that the jar is written, once an entry
itself has been written, we know that all of its parent directories
will also have been written.
This commit updates the jar writer so that no attempt is made to
write parent directory entries if the entry itself has already been
written.
Fixes gh-29175
