Fix mp.jwt.verify.publickey check in SmallryeJwtDevModeProcessor

[rsvoboda]

Fix mp.jwt.verify.publickey check in SmallryeJwtDevModeProcessor

Fixes #46928 / native mode case

[rsvoboda]

oki, I will fix this. it's (SmallryeJwtDevModeProcessor) obviously also for tests

[gsmet]

I would expect the native build to run in normal mode? So I'm a bit surprised it makes a difference?

[rsvoboda]

my bet is NativeSourcesBuild part, but I can look more into it

[rsvoboda]

Having just @BuildStep(onlyIfNot = { IsNormal.class, NativeSourcesBuild.class }) doesn't fix the issue

So my change is the minimal one.

[gsmet]

I'm not exactly excited about this change and I would like us to get to the bottom of it.

What I understand is that you are pointing to a non-existing public key and hoping things will work.

Also I'm not exactly sure why you have JWT around but jwt is not really used.

Could you clarify exactly what you're doing in this test, why you need to define a JWT key that points to nowhere - and why you expect this to be fine.

Maybe your patch is right but at this point I can't be sure.

[rsvoboda]

BeefyTS is legacy and we keep it alive to see if old approaches work. The 003-quarkus-many-extensions module combines a lot of extensions and the intention is to ensure it can be built in JVM and Native mode. JWT properties in https://github.com/quarkus-qe/beefy-scenarios/blob/main/003-quarkus-many-extensions/src/main/resources/application.properties#L4-L5 were needed when the app was composed, now the JWT extension doesn't need these properties to be defined when no JWT logic is triggered. I can remove them and things will work again.

I identified the PR that caused failure of the module in native. It's #44272 as mentioned in #46928 (comment)

Very strange thing is that JVM mode doesn't fail, but native mode fails trying to read the /foo/bar from mp.jwt.verify.publickey.location property.

#44272 states that Generate RSA-256 keys on dev mode and that got me slightly sidetracked, it's apparently for dev mode and tests. So the title of PR should be renamed and also maybe alsoSmallryeJwtDevModeProcessor.java should be renamed to indicate also test, e.g. SmallryeJwtDevTestModeProcessor.java.

So in the end I think I will just adjust 003-quarkus-many-extensions and drop the mp.jwt dummy config.

[rsvoboda]

oh, I think I hit something :) I will update the PR

[rsvoboda]

@gsmet PR updated to check userConfigs for mp.jwt.verify.publickey, it also fixes my old/weird case

[rsvoboda]

I have updated #44272 title to mention also test

Let me know about the SmallryeJwtDevModeProcessor file naming. Should I also rename it to SmallryeJwtDevTestModeProcessor or I'm just being too choosey :) ?

[sberyozkin]

Thanks @rsvoboda 🙏

[quarkus-bot]

Status for workflow Quarkus CI

This is the status report for running Quarkus CI on commit 1675e64.

✅ The latest workflow run for the pull request has completed successfully.

It should be safe to merge provided you have a look at the other checks in the summary.

You can consult the Develocity build scans.

[sberyozkin]

As far as I understand, this fix from @rsvoboda fixes the typo in the @mcruzdev's PR, @mcruzdev can you please check?

[sberyozkin]

I'll hold off the merge for now, I'm on the road and missed there was the discussion, I think the fix is good, but would appreciate @mcruzdev having a look.

[mcruzdev]

I am taking a look now

[mcruzdev]

If the user does not set the property then should be setted as NONE. Thanks a lot @rsvoboda, and sorry guys!

@sberyozkin, @gsmet I think we can merge this one.

[sberyozkin]

Thanks @mcruzdev

Latest PR update fixes the typo, with the rationale behind this update also clarified. #44272 has tests but I think the ordering of checks have prevented detecting the typo earlier