cli: idl close to directly an idl address

[ochaloup]

On upgrading the IDL account

anchor --provider.cluster devnet idl \
  --provider.wallet <fee-payer> upgrade <program-id> -f ./target/idl/program.json

It creates the IDL buffer and then it writes the buffer to the program IDL account.
The IDL buffer is still opened. It's not possible to use anchor idl close with specific IDL address (e.g., address of the buffer) to close it and get back the rent for the account.
Adding the optional argument of providing the IDL address directly makes possible to close the buffer account as well.

[vercel]

@ochaloup is attempting to deploy a commit to the coral-xyz Team on Vercel.

A member of the Team first needs to authorize it.

[acheroncrypto]

I wonder why upgrade command doesn't close the buffer account by default, perhaps to version the IDLs? If you don't have any objections, I think we should close the buffer account similar to how program upgrades work and historical data should be left to RPC providers.

[ochaloup]

I wonder why upgrade command doesn't close the buffer account by default, perhaps to version the IDLs? ...

To me the suggestion makes perfect sense. I would expect it should work this way to follow how the solana program cli does.
Do you thing adding the idl_close call at id_upgrade makes this working as you proposed?
https://github.com/coral-xyz/anchor/pull/2760/files#diff-c1f8f7498da827a634bddc8a7559198bc99b296e9d9e8b91a70b503662995b8cR2264

[acheroncrypto]

Do you thing adding the idl_close call at id_upgrade makes this working as you proposed?

Yes, it works great but the upgrade output is a bit confusing:

Idl buffer created: CK5yrU9hjfgPRjEN5PKTLK7UD6niW1C7RvBXstrwjJfm
Idl account closed: CK5yrU9hjfgPRjEN5PKTLK7UD6niW1C7RvBXstrwjJfm

Maybe we should log the upgrade was successful and the IDL account key?

Also, could you note both of these changes in the CHANGELOG? I think closing the buffer account is a breaking change since it's possible some people are using the buffer account.

[ochaloup]

@acheroncrypto Sure. So it's a new update here in PR. Let's see.

[acheroncrypto]

Somehow the previous change that removed short was overridden but I restored it. Thanks!

[ochaloup]

Somehow the previous change that removed short was overridden but I restored it. Thanks!

Ah, that was probably my wrong, sorry.

Thanks for looking into this, for fixing and merging 😉