NeXeNyou could use @ServiceConnection and make a singleton .....i like using flyway schema reconstruction to cleanup between test executions
NeXeNyou could also use specific names or tags so the static registry produces distinct container entries
NeXeNyou know, if it were strictly necessary
deeboit's not, we're mostly trying to keep everything as "springy" as possible without test specific context configurations
NeXeNyou could do like a zero config test isolation with @Transactional .....roll back after test so no state leakage
deebothat's what we do, but there's 10+ years of history, lots of different types of tests on different levels
NeXeNyou could do property driven dynamic urls with TC_INITSCRIPT.....target them directly in application-test.properties or in the yaml if you use it
NeXeNTestExecutionListener? maybe instead of decorating every class just use the SpringLoaderFactory SPI
NeXeNwait no....SpringFactoriesLoader?
NeXeNdamn i'm slydexic
dreamrealOr you could do what I did :D
NeXeNbut any @SpringBootTest across the code base would then automagically run the listener without changing code. an invisible global lifecycle listener. you could trunc tables after exectuion then toss it in your spring.factories
dreamreal... I mean, I included source code
NeXeNwhat version of spring did you say? nothing 3.x yet ?
NeXeNi have never had to worry about a large history of tests like that, but it seems no dev would need to touch a thing really. as long as they're already annotated
dreamrealIt was 4
dreamrealbut it wouldn't be a ginormous effort to do it on 3
* NeXeN ponders
dreamrealwoo, it's back! It's muted here, but it's aliiiiiive
DoofusCanadensiscue Igor
dreamrealI should have named it igor, or eyegore, those are good names
dreamreal(enigmastation.com is my blog; bytecode.news is a general news site.)
dreamrealThat FSF thing feels like an early writer to me, mostly because it was. The 2014 date is when I captured the copy; the original writing would probably have been 1999 or so.