The try/catch syntax introduced in 0.6.0 is arguably the largest leap in error dealing with capabilities in Solidity, since cause strings for revert and require have been launched in v0.4.22. Each attempt and catch have been reserved key phrases since v0.5.9 and now we are able to use them to deal with failures in exterior operate calls with out rolling again the entire transaction (state modifications within the referred to as operate are nonetheless rolled again, however the ones within the calling operate usually are not).
We’re shifting one step away from the purist “all-or-nothing” strategy in a transaction lifecycle, which falls wanting sensible behaviour we frequently need.
Dealing with exterior name failures
The attempt/catch assertion lets you react on failed exterior calls and contract creation calls, so you can’t use it for inside operate calls. Observe that to wrap a public operate name throughout the similar contract with attempt/catch, it may be made exterior by calling the operate with this..
The instance under demonstrates how attempt/catch is utilized in a manufacturing facility sample the place contract creation would possibly fail. The next CharitySplitter contract requires a compulsory deal with property _owner in its constructor.
pragma solidity ^0.6.1; contract CharitySplitter { deal with public proprietor; constructor (deal with _owner) public { require(_owner != deal with(0), "no-owner-provided"); proprietor = _owner; } }
There’s a manufacturing facility contract — CharitySplitterFactory which is used to create and handle cases of CharitySplitter. Within the manufacturing facility we are able to wrap the new CharitySplitter(charityOwner) in a attempt/catch as a failsafe for when that constructor would possibly fail due to an empty charityOwner being handed.
pragma solidity ^0.6.1; import "./CharitySplitter.sol"; contract CharitySplitterFactory { mapping (deal with => CharitySplitter) public charitySplitters; uint public errorCount; occasion ErrorHandled(string cause); occasion ErrorNotHandled(bytes cause); operate createCharitySplitter(deal with charityOwner) public { attempt new CharitySplitter(charityOwner) returns (CharitySplitter newCharitySplitter) { charitySplitters[msg.sender] = newCharitySplitter; } catch { errorCount++; } } }
Observe that with attempt/catch, solely exceptions occurring contained in the exterior name itself are caught. Errors contained in the expression usually are not caught, for instance if the enter parameter for the new CharitySplitter is itself a part of an inside name, any errors it raises won’t be caught. Pattern demonstrating this behaviour is the modified createCharitySplitter operate. Right here the CharitySplitter constructor enter parameter is retrieved dynamically from one other operate — getCharityOwner. If that operate reverts, on this instance with “revert-required-for-testing”, that won’t be caught within the attempt/catch assertion.
operate createCharitySplitter(deal with _charityOwner) public { attempt new CharitySplitter(getCharityOwner(_charityOwner, false)) returns (CharitySplitter newCharitySplitter) { charitySplitters[msg.sender] = newCharitySplitter; } catch (bytes reminiscence cause) { ... } } operate getCharityOwner(deal with _charityOwner, bool _toPass) inside returns (deal with) { require(_toPass, "revert-required-for-testing"); return _charityOwner; }
Retrieving the error message
We are able to additional lengthen the attempt/catch logic within the createCharitySplitter operate to retrieve the error message if one was emitted by a failing revert or require and emit it in an occasion. There are two methods to realize this:
1. Utilizing catch Error(string reminiscence cause)
operate createCharitySplitter(deal with _charityOwner) public { attempt new CharitySplitter(_charityOwner) returns (CharitySplitter newCharitySplitter) { charitySplitters[msg.sender] = newCharitySplitter; } catch Error(string reminiscence cause) { errorCount++; CharitySplitter newCharitySplitter = new CharitySplitter(msg.sender); charitySplitters[msg.sender] = newCharitySplitter; // Emitting the error in occasion emit ErrorHandled(cause); } catch { errorCount++; } }
Which emits the next occasion on a failed constructor require error:
CharitySplitterFactory.ErrorHandled( cause: 'no-owner-provided' (sort: string) )
2. Utilizing catch (bytes reminiscence cause)
operate createCharitySplitter(deal with charityOwner) public { attempt new CharitySplitter(charityOwner) returns (CharitySplitter newCharitySplitter) { charitySplitters[msg.sender] = newCharitySplitter; } catch (bytes reminiscence cause) { errorCount++; emit ErrorNotHandled(cause); } }
Which emits the next occasion on a failed constructor require error:
CharitySplitterFactory.ErrorNotHandled( cause: hex'08c379a0000000000000000000000000000000000000000000000000000000000000002000000000000000000000000000000000000000000000000000000000000000116e6f2d6f776e65722d70726f7669646564000000000000000000000000000000' (sort: bytes)
The above two strategies for retrieving the error string produce the same end result. The distinction is that the second technique doesn’t ABI-decode the error string. The benefit of the second technique is that it is usually executed if ABI decoding the error string fails or if no cause was supplied.
Future plans
There are plans to launch help for error varieties that means we can declare errors in the same solution to occasions permitting us to catch completely different sort of errors, for instance:
catch CustomErrorA(uint data1) { … } catch CustomErrorB(uint[] reminiscence data2) { … } catch {}