Juraj Kirchheim

Yeah, as I said we should probably pass the whole request. I guess the only question is whether we want to retain support for session implementations that consume the header...

Seems to work fine now.

Oh ... maybe not ^^

Added bonus: if `-main` class has no main method, `serve` it.

I don't really know how the cache invalidation works, but `tink.macro.BuildCache` does manage to deal with the problem: https://github.com/haxetink/tink_macro/blob/master/src/tink/macro/BuildCache.hx#L180

Uhm, technically that should be a 422 (UnprocessableEntity). If it's not, then that's a bug in tink_json or possibly a bug in tink_web swallowing the error code. If it is,...

The problem with this is that tink_xml structures are typically underspecified (or rather: XML quite often contains significant redundancy). Let's summon @piboistudios, as has been working on that quite recently....

Well, the main problem here is parsing `DynamicAccess`, which is currently unsupported. It'll require restructuring [how the data is stored in the parser](https://github.com/haxetink/tink_querystring/blob/f1710c07d8ed5d01521c3db10288f7c9733c0411/src/tink/querystring/Parser.hx#L12), because getting maps from a flat key-value-list...

Yeah, but headers are basically a key-value list too ... both get parsed via tink_querystring, which despite its name is not *that* query string specific ^^

For the query it should work (in theory). For the body, this is is pretty much intended, because we don't want to begin body parsing before access is granted. But...