Removal of default parameter is binary compatible but breaks runtime about binary-compatibility-validator

[fzhinkin]

• Should it be implemented only for new Klib format?
• It will be a breaking change so should it be done before Klib finaly released?

In fact, Klib format already tracks parameters with default arguments, so it should be added only for JVM dumps. It indeed will be a breaking change, but the severity of the problem it solves justifies such a change.

Anyway, we can make a new behavior optional and, a few releases later, make it default.

• Do you have an idea on how it could be done? Will it be some mark on each parameter having default value, or only the bitmask parameter will have some additional info?

I was thinking about printing all extra metadata near an affected declaration, but there is no particular proposal so far.

• Do you think this change could be contributed by someone who is not familiar with the codebase? If so, I want to try to do it.

Sure! Feel free to implement and contribute it; let us just finalize how all the extra info should be presented in the dump first (I hope it'll happen within a few weeks).

[fzhinkin]

Sorry for the late reply. Indeed, it does make sense to capture information about parameters with default arguments in a generated dump.

[osipxd]

Great! I have the following questions:

• Should it be implemented only for new Klib format?
• It will be a breaking change so should it be done before Klib finaly released?
• Do you have an idea on how it could be done? Will it be some mark on each parameter having default value, or only the bitmask parameter will have some additional info?
• Do you think this change could be contributed by someone who is not familiar with the codebase? If so, I want to try to do it.