eip165 - Pseudo-Introspection, or standard interface detection #639

jbaylina · 2017-05-26T11:23:22Z

Formal PR for Pseudo-Introspection of #165

Arachnid

Agh, it seems I never submitted this review.

Arachnid · 2017-05-27T13:19:08Z

EIPS/eip-165.md

+
+##### Procedure to determine if a contract implements EIP165 standard.
+
+1. The source contact makes a `CALL` to the destination address with input data: 0x01ffc9a701ffc9a7 value: 0 and gas 30000


This should properly be padded to the right with zeroes, but I think it would be clearer to annotate this at function level - eg "the source contract calls supportsInterface(0x01ffc9a7)".

Arachnid · 2017-05-27T13:20:04Z

EIPS/eip-165.md

+
+1. The source contact makes a `CALL` to the destination address with input data: 0x01ffc9a701ffc9a7 value: 0 and gas 30000
+2. If the call fails or return false, the destination contract does not implement EIP165
+3. If the call returns true, a second call is made with input data: 0x01ffc9a7ffffffff


Is this necessary? The chances of a contract randomly returning true to everything seem pretty low.

Also, I suspect that in most cases we know the target contract implements supportsInterface, but not which interfaces it implements - such as is the case with ENS resolvers, for instance.

A typical example could be to ask if a wallet implements ITokenFallback. In most cases, this call will throw or return false. The return true case, I see it very unprovable now, but after metropolis I see the possibility that some day solidity could be able to return values in the fallback function. (This could be very useful for implementing proxy contracts for example). I also do not control other languages like serpent or direct assembly contracts. So because of this possibility and to be sure there is no edge cases, I implemented the "Double call", But if you don't see it probable, we can remove this double check and remve point 3.

Personally I don't think it's necessary.

Arachnid · 2017-05-27T13:20:49Z

EIPS/eip-165.md

+
+Returns true if `_contract` supports EIP165
+
+    function interfaceSupported(address _contract, bytes4 _interfaceId) constant returns (bool);


Checking if the contract implements EIP165 aside, will this be any more efficient than just asking the contract directly?

If the contract throws (because it is an old one) it will consume all the gas.... So in this case I suspect that it's much more efficient to have a cache. For a new contract (and if we remove the double call), probably it will not be very different.

Good point on the throw.

If we assume that we will use a cache, then may be it's not so important to remove the "double call" in terms of performance (It will be done only once by the cache contract). And this way we don't risk for edge cases.

True; how about rewording it such that it's optional? In cases where you already know the target contract implements supportsInterface, there's no point in calling it more than once.

jbaylina · 2017-11-15T15:27:55Z

Please check #672 as alternative to this EIP.

Arachnid · 2017-11-15T16:28:34Z

I'd still like to standardise this. ENS can't use #672 without massive ouroboros confusion. ;)

Arachnid · 2017-11-16T12:44:00Z

EIPS/eip-165.md

+
+##### Procedure to determine if a contract implements EIP165 standard.
+
+1. The source contact makes a `CALL` to the destination address with input data: 0x01ffc9a701ffc9a7 value: 0 and gas 30000. `0x01ffc9a701ffc9a7` is the data for `supportsInterface(0x01ffc9a7)`. Asks if self implements itself.


I still think this needs to be expressed as an ABI, not as raw bytes.

The problem is that in solidity you cannot catch the throw inside a call by just calling in the ABI way. You will have to use the call by hand and check the result. I added the explanation in the same line, but I think is worthy to just explain how the internal call is make. Please fill free to rewrite this paragraph in a way you fill more confortable.

Arachnid · 2017-11-16T12:46:35Z

EIPS/eip-165.md

@@ -0,0 +1,229 @@
+## Preamble
+
+    EIP: <to be assigned>


This needs setting.

fulldecent · 2018-01-20T05:21:03Z

This function should be pure if caching will work!

fulldecent · 2018-01-20T05:23:15Z

^^ @RuFeDoRu @Arachnid

fulldecent · 2018-01-21T20:03:23Z

I don't like how the contract needs to know about the cache. This is surely more complicated than need be and the control flow is inverted.

Just make things simple!

fulldecent · 2018-01-21T20:06:01Z

The specification need not be any more complicated than this:

pragma solidity ^0.4.19;

/// @title Interface for contracts conforming to ERC-165: Standard Interface Detection
/// @author XXX, William Entriken (https://phor.net)
/// @dev Draft specification at https://github.com/ethereum/eips/issues/165
/// @dev Limitations in Solidity prevent us from declaring this as an interface
/* interface */ contract ERC165 {
    /// @dev The way to calculate an interface's ID is documented in the ERC
    ///  But we also provide it here for convenience.
    bytes4 internal constant INTERFACE_SIGNATURE_ERC165 = // 0x01ffc9a7
        bytes4(keccak256('supportsInterface(bytes4)'));

    /// @notice Query a contract to see if it supports a given interface
    /// @dev See ERC-165 documentation on how to calculate `_interfaceID`
    /// @param _interfaceID The selected interface to test
    /// @return true if the interface is supported
    function supportsInterface(bytes4 _interfaceID) external pure returns (bool);
}

Everything else is just delaying review and stalling this from becoming an actual standard.

nicksavers · 2018-02-18T17:19:57Z

@jbaylina Is this now superseded by #881 (ERC-165 Standard Interface Detection) and do you support it?

fulldecent · 2018-02-19T19:29:34Z

I reject my comment about pure being necessary. We have updated the ERC165Cache discussed here. Updated version at #881. We hope you will endorse this as the canonical pull request for the ERC #165.

jbaylina added 2 commits May 26, 2017 13:17

eip165

d8a3b35

Small typos and language fixes

1cac64c

RuFeDoRu approved these changes May 28, 2017

View reviewed changes

Arachnid reviewed Nov 15, 2017

View reviewed changes

Nick comments added

d33102e

Arachnid approved these changes Nov 16, 2017

View reviewed changes

Arachnid reviewed Nov 16, 2017

View reviewed changes

EIPS/eip-165.md Outdated

@@ -0,0 +1,229 @@

## Preamble

EIP: <to be assigned>

Copy link

Contributor

Arachnid Nov 16, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This needs setting.

EIP number assigned

4a9c25c

pirapira added the ERC label Dec 1, 2017

fulldecent mentioned this pull request Jan 19, 2018

ERC: Non-fungible Token Standard #721

Closed

fulldecent mentioned this pull request Jan 20, 2018

Pseudo-Introspection, or standard interface detection #165

Closed

ethereum deleted a comment from RuFeDoRu Jan 20, 2018

fulldecent mentioned this pull request Feb 19, 2018

ERC777 Token Standard #777

Closed

jbaylina closed this Feb 20, 2018

fulldecent mentioned this pull request Feb 20, 2018

ERC-165 Standard Interface Detection #881

Merged

8 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

eip165 - Pseudo-Introspection, or standard interface detection #639

eip165 - Pseudo-Introspection, or standard interface detection #639

jbaylina commented May 26, 2017

Arachnid left a comment

Arachnid May 27, 2017

Arachnid May 27, 2017

jbaylina Nov 15, 2017 •

edited

Loading

Arachnid Nov 15, 2017

Arachnid May 27, 2017

jbaylina Nov 15, 2017

Arachnid Nov 15, 2017

jbaylina Nov 15, 2017

Arachnid Nov 15, 2017

jbaylina commented Nov 15, 2017

Arachnid commented Nov 15, 2017

Arachnid Nov 16, 2017

jbaylina Nov 16, 2017

Arachnid Nov 16, 2017

fulldecent commented Jan 20, 2018

fulldecent commented Jan 20, 2018

fulldecent commented Jan 21, 2018

fulldecent commented Jan 21, 2018

nicksavers commented Feb 18, 2018

fulldecent commented Feb 19, 2018


		##### Procedure to determine if a contract implements EIP165 standard.

		1. The source contact makes a `CALL` to the destination address with input data: 0x01ffc9a701ffc9a7 value: 0 and gas 30000


		Returns true if `_contract` supports EIP165

		function interfaceSupported(address _contract, bytes4 _interfaceId) constant returns (bool);

eip165 - Pseudo-Introspection, or standard interface detection #639

eip165 - Pseudo-Introspection, or standard interface detection #639

Conversation

jbaylina commented May 26, 2017

Arachnid left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jbaylina Nov 15, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jbaylina commented Nov 15, 2017

Arachnid commented Nov 15, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

fulldecent commented Jan 20, 2018

fulldecent commented Jan 20, 2018

fulldecent commented Jan 21, 2018

fulldecent commented Jan 21, 2018

nicksavers commented Feb 18, 2018

fulldecent commented Feb 19, 2018

jbaylina Nov 15, 2017 •

edited

Loading