Martin Kouba
Martin Kouba
> Damnit, I know from the start that the syntax should have been `#{include 'foo.html'/}` 🤦 Yup, we make mistakes, we learn... but we don't want to break existing stuff...
CC @Ladicek
> I believe it makes most sense to create an intermediate representation that describes what code should be generated, and only after that is constructed, do the generation. @Ladicek Could...
> That said, maybe we could examine some specific existing cases and see how this theory looks in practice. Yes! I'd love to do that once we get past the...
> We could determine that info beforehands, in an object called e.g. ClientProxyCodeGenInfo, and just follow the info in that object when doing codegen. Ok, I understand! Yes, this would...
> How are we feeling about this, three weeks on? Is this something that we can resolve in documentation (for example)? Yes, we could but I would still keep this...
> As I mentioned, I started porting ArC to Gizmo 2. I naturally started with the simplest cases, but got far enough to have a first-hand experience with this, see...
FYI this one - https://github.com/quarkusio/quarkus/pull/51396 - could help to identify the build steps that produce a large number of `Reflective*BuildItem`s.
A template root is always relative to an application artifact, or the root dir. It cannot be an arbitrary directory because the scanning logic must work in all _run modes_...
The HR [`Mutiny.Session` bean was removed](https://github.com/quarkusio/quarkus/pull/31121) after the [big refactoring of HR Panache](https://github.com/quarkusio/quarkus/pull/29761). We had many problems with concurrency access, session leaks, context propagation, etc. Note that at that time...