BUGFIX: Implement document uri path placeholders

[nezaniel]

resolves #5077

This introduces the concept of placeholders for document uri paths. If the DocumentUriPathProjection encounters an unknown node type - which may happen during replay - it creates a placeholder instead of ignoring it. Placeholders are not evaluated in routing, but are transformed to actual uri paths once the respective node aggregate has been changed to type document, shortcut or site.

Upgrade instructions

needs both cr:setup and cr:projectionreplay documenturipath

Review instructions

We still need to decide on how to deal with the node type changes

Checklist

• Code follows the PSR-2 coding style
• Tests have been created, run and adjusted as needed
• The PR is created against the 9.0 branch
• Reviewer - PR Title is brief but complete and starts with FEATURE|TASK|BUGFIX
• Reviewer - The first section explains the change briefly for change-logs
• Reviewer - Breaking Changes are marked with !!! and have upgrade-instructions

[bwaidelich]

Clever solution, +1 by reading.
Just 2 nitpicks, potentially as conversation starter – feel free to ignore.

One thing I wonder: IIRC we plan to store the expanded super node type names with the creation event. That might make parts of this redundant!?

[nezaniel]

It will prevent future events from having classification unknown, yes. But sadly we cannot safely apply the SuperNodeTypes to past events because we don't know the past state anymore, so we still need to handle that case

[mhsdesign]

But sadly we cannot safely apply the SuperNodeTypes to past events because we don't know the past state anymore, so we still need to handle that case

i would like to think further into that direction if its really not possible as this will prevent us from this complexity. But for that i have think a little more :D

[nezaniel]

Proof that this cannot work:

• have NodeTypes A and B with supertype Document
• event 1: Create a node of type A
• event 2: Change type to B
• event 3: Remove type A

Now try to assign supertypes to event 1

[nezaniel]

Also yet tbd: what to do if something is a site AND a shortcut

[grebaldi]

Hi @nezaniel,

I was able to verify that your change resolves #5077.

I have one (unfortunately non-trivial) note on the code (see comments below), other than that: 👍

Regarding that note: I'm hesitant to select "Request Changes" here, because my suggestion is in a gray area between nice-to-have and probably-better-for-long-term-comprehensibility, and will require quite some work. I can't judge how relevant that is in the context of this particular projection. So, I'd be fine if you 👎 me on this one. I'll approve regardless of my suggestion in that case.

[grebaldi]

@nezaniel and I talked about my comments above and concluded that these are possible improvements that can be taken on at a later point, and should be postponed to enable a swift beta11 release.

To ensure that the current state is not exposed as API anywhere, @nezaniel has now marked DocumentTypeClassification as @internal.