Use the path parameter of encode/decode again to be future-proof

[syssi]

cp. #220 (comment)

[coveralls]

Coverage remained the same at 67.951% when pulling 74e9bed on syssi:feature/construct-encode-decode-path into 26fc798 on rytilahti:master.

[coveralls]

Coverage remained the same at 67.951% when pulling 74e9bed on syssi:feature/construct-encode-decode-path into 26fc798 on rytilahti:master.

[coveralls]

Coverage remained the same at 67.951% when pulling 74e9bed on syssi:feature/construct-encode-decode-path into 26fc798 on rytilahti:master.

[rytilahti]

I suppose the question to ask here is whether we are more afraid of potential silent errors (which we can handle by checking against incoming kwargs) or changes in the interface used by construct. Maybe keeping the path as it is, and adding **kwargs in case the interface gets extended at some point, to avoid crashing it all immediately?

[yawor]

There's still an open issue #224 when using Construct > 2.9.28.

[arekbulski]

Instead of crashing "immediately" due to error explictly related to Adapter code, it will fail "later" due to unknown reasons that dont flag construct lines of code in any way. You might as well wrap you entire code in try/catch and ship it.