Changing ListMetaData to correspond to reality

[erikbosch]

We do not use the filter field, and the name root is not that logical. In older API supporting same wildcard syntax we use the term path, so proposing to change to that.

https://github.com/eclipse-kuksa/kuksa-databroker/blob/main/proto/kuksa/val/v1/val.proto#L104

message SubscribeEntry {
  string path           = 1;
  View view             = 2;
  repeated Field fields = 3;
}

Related to discussion at https://github.com/eclipse-kuksa/kuksa-databroker/pull/105/files#r1848167147

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 59.34%. Comparing base (66b1c91) to head (21c7bd0).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #106      +/-   ##
==========================================
- Coverage   59.38%   59.34%   -0.05%     
==========================================
  Files          33       33              
  Lines       16048    16043       -5     
==========================================
- Hits         9530     9520      -10     
- Misses       6518     6523       +5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

---

🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests

[erikbosch]

There is ongoing discussion if we better should change implementation to reflect the *.proto file. I.e. have

• root - that must be a path to branch without wildcards.
  • OK Examples Vehicle, Vehicle.Cabin.
  • Bad Examples: Vehicle*, Vehicle.Speed, *
• filter: Path possibly including wildcard

Below some possible examples to show results for a spec containing

Branch A
  Branch A.B
    Sensor A.B.C
    Sensor A.B.CC
    Sensor A.B.D
  Branch A.BB
    Sensor A.BB.A
  Branch A.C // Empty branch

Root	Filter	Result
A	None	A.B.C, A.B.CC, A.B.D and A.BB.A
A	""	A.B.C, A.B.CC, A.B.D and A.BB.A
A	*	A.B.C, A.B.CC, A.B.D and A.BB.A Or should * only match a single level to differentatiate from **
A.B	None	A.B.C, A.B.CC and A.B.D
A.B	""	A.B.C, A.B.CC and A.B.D
A.B	*	A.B.C, A.B.CC and A.B.D
A.B	C	A.B.C Only!
A.B	C*	A.B.C and A.B.CC
A.B*	None	Error - wildcard not allowed in root - INVALID_ARGUMENT
A.B.C	None	Error - root must not be a signal! - INVALID_ARGUMENT
A.C	None	Empty list returned as branch exist but is empty (if we can differentiate the case from the next)
A.D	None	Error - Branch does not exist - NOT_FOUND
A.B	Sebastian	Empty list returned as signal A.B.Sebastian does not exist

FYI @rafaeling

@argerus - does the table match how you think it should work

[erikbosch]

Put on hold for now, we possibly want to have root plus filter anyway, so at least no proto update for 5.0

[BjoernAtBosch]

Some points of critic:

• With the removed filter one cannot request certain "types" of metadata any longer (like "I just want to get the numeric ids"). This may result in a lot of (unneeded) data.
• Having just one path (formerly root), it's hard to request metadata in a single request for completely different branches, like Vehicle.Speed and Vehicle.Cabin.Seat.*. (Or?)

[erikbosch]

Well as of today we do not care about "filter" at all. I have no problem having a filter field if we actually intend to use it, but we do not describe how filter shall/can be used at all right now, like if it going to support the use-cases you mention. But as we seem to be uncertain I put this PR on hold.

In #105 I instead pointed out that filter currently is not used, and that current implementation may change. Like to be on the safe side, do not use wildcards in root, as it may not be supported in the future.

[BjoernAtBosch]

(Don't know, where to put this finding else:)
Rename field datapoints of message PublishValuesRequest to data_points(w/ underscore).