Martin Kouba

> I'm actually not sure @ParameterizedTests are well supported in @QuarkusTest. It is not. We will have to revisit the `io.quarkus.test.TestMethodInvoker` SPI because currently only one invoker is supported and...

> [@mkouba](https://github.com/mkouba) should we close all these issues as `won't fix`, or do you believe there is a near future hope. (Closing does not mean it won't be done, just...

> > Is there an easy way in the RedisClientProcessor class that we could query dev services to determine if it has initialized any named beans/default beans for a class...

> The desired behavior is, if hosts null or empty AND dev services not running for this instance, fail to start, else start. I think that I understand now. Thanks...

> > I'm not quite sure how to solve this problem in general. [@cescoffier](https://github.com/cescoffier) [@radcortez](https://github.com/radcortez) > > This is expected. Setting an empty value unsets the configuration. Configuration always worked...

The failures are flaky tests, I will merge this one.

> I guess it might have been a mis-usage of the API ? Is there another way to pass more than one .properties resource per QuarkusUnitTest ? Well, the contract...

> Thanks. If `#withConfigurationResource` behavior will remain as-is and it's a new method, I guess short-term I'll refactor my tests anyway to use `#overrideConfigKey()` or merge the .properties, as I...

> I would be very much in favor of removing `withConfigurationResource` and `overrideConfigKey`. 👎 for removing `overrideConfigKey()`, it is a very convenient and transparent way to set a config property...

Given the fact that all those "overriden" config properties end up in the same config source (`application.properties`) I don't think it's a problem to add one more method to `QuarkusUnitTest`...