[Move devex] Move code discovery (umbrella task)

[awelc]

This is an umbrella task, so please don't (self)assign this task as a whole but rather created sub-tasks and assign those (different subtasks can be matched with different timeline targets).

We should have some way of discovering what type of code is available on-chain so that developers can re-use this code if they wish. Some ideas include:

• add package search for the object explorer (not sure how to do a meaningful search without disassembling stored bytecodes which is relatively costly)
• make a connection between source code and on-chain bytecode in an off-chain tool (perhaps make it part of Movey?)
• store source-code on chain and make (chain-verifiable) connection between bytecode and source code (possibly in an anonymous fashion where source code for a given on-chain bytecode is submitted by someone, we compile the source code, and make the connection only if compiled bytecode matches on-chain bytecode - this is how it's done in Eth).

SUBTASKS:

• [ ] https://github.com/MystenLabs/sui/issues/1115

[tnowacki]

make a connection between source code and on-chain bytecode in an off-chain tool

We had talked a while ago about adding an annotation/metadata section to the compiled module that would jsut be used by the adapter.

One field we might want to add is a hash of the source project. I think this + the ability to recompile the code is all you need to associate source code with published code.

store source-code on chain

You probably dont want to fully store the code on chain? Seems like a lot of data

[awelc]

store source-code on chain

You probably dont want to fully store the code on chain? Seems like a lot of data

If I understand it correctly:

• a Move code developer puts bytecode on-chain with a hash code of their own source code used to compile the program
• the same Move code developer (though it's not a strict requirement I guess) publishes source code off-chain
• everyone else that wants to make sure that the the two match computes the hash of the off-chain source code and compares it with the hash stored on-chain

Is this the intended flow?

[tnowacki]

store source-code on chain

You probably dont want to fully store the code on chain? Seems like a lot of data

If I understand it correctly:

• a Move code developer puts bytecode on-chain with a hash code of their own source code used to compile the program
• the same Move code developer (though it's not a strict requirement I guess) publishes source code off-chain
• everyone else that wants to make sure that the the two match computes the hash of the off-chain source code and compares it with the hash stored on-chain

Is this the intended flow?

That is what I had in mind, yes.

But I think you need to change that last step to:

• everyone else that wants to make sure that the the two match:
  • computes the hash of the off-chain source code and compares it with the hash stored on-chain
  • compiles the source code and makes sure it matches the compiled module on chain
  • this is tricky with compiler versioning so not sure if there is something better we can do here