Change Details

The "core" of the interpreter is `IguanaInterpreter::Run`, which aims to exactly model the VerifyScript function. The internal RunScript function is modelled after `EvalScript`, but it collects a detailled trace of every opcode, i.e. the state of the stack and altstack. It also collects errors in an as helpful as possible way. For example, integer overflows (which bubble up as exceptions in EvalScript), now finally have a nice error message (bitcoind would report "Unknown error"). For a scriptSig, it will stop at the first non-push opcode, instead of just saying that it has some non-pushop somewhere. And other cases, e.g. invalidly encoded opcodes. After collecting the result, a Formatter can be used to display the trace. One implementation is `FormatterHumanReadable`, which prints it to stdout in a somewhat verbose but easy to understand format. The executable itself simply takes `-tx`, `-scriptpubkey`, `-value` and `-inputindex`, and runs the interpreter. This is similar to the parameters Meep takes, but named after the fields in the bitcoind codebase. In the future, we can add a `-format` arg, which could allow CSV or other machine readable formats (that could be fed into Script IDEs, e.g. Bitauth IDE).

The "core" of the interpreter is `IguanaInterpreter::Run`, which aims to exactly model the VerifyScript function. The internal RunScript function is modelled after `EvalScript`, but it collects a detailled trace of every opcode, i.e. the state of the stack and altstack. It also collects errors in an as helpful as possible way. For example, integer overflows (which bubble up as exceptions in EvalScript), now finally have a nice error message (bitcoind would report "Unknown error"). For a scriptSig, it will stop at the first non-push opcode, instead of just saying that it has some non-pushop somewhere. And other cases, e.g. invalidly encoded opcodes. After collecting the result, a Formatter can be used to display the trace. One implementation is `FormatterHumanReadable`, which prints it to stdout in a somewhat verbose but easy to understand format. The executable itself simply takes `-tx`, `-scriptpubkey`, `-value` and `-inputindex`, and runs the interpreter. This is similar to the parameters Meep takes, but named after the fields in the bitcoind codebase. In the future, we can add a `-format` arg, which could allow CSV or other machine readable formats (that could be fed into Script IDEs, e.g. Bitauth IDE). Also, we can add a `-scriptflags` arg, to allow e.g. standard vs. mandatory flags, or a specific custom flagset. This would also allow us to run the debugger against the script_tests.json.

The "core" of the interpreter is `IguanaInterpreter::Run`, which aims to exactly model the VerifyScript function. The internal RunScript function is modelled after `EvalScript`, but it collects a detailled trace of every opcode, i.e. the state of the stack and altstack. It also collects errors in an as helpful as possible way. For example, integer overflows (which bubble up as exceptions in EvalScript), now finally have a nice error message (bitcoind would report "Unknown error"). For a scriptSig, it will stop at the first non-push opcode, instead of just saying that it has some non-pushop somewhere. And other cases, e.g. invalidly encoded opcodes. After collecting the result, a Formatter can be used to display the trace. One implementation is `FormatterHumanReadable`, which prints it to stdout in a somewhat verbose but easy to understand format. The executable itself simply takes `-tx`, `-scriptpubkey`, `-value` and `-inputindex`, and runs the interpreter. This is similar to the parameters Meep takes, but named after the fields in the bitcoind codebase. In the future, we can add a `-format` arg, which could allow CSV or other machine readable formats (that could be fed into Script IDEs, e.g. Bitauth IDE). Also, we can add a `-scriptflags` arg, to allow e.g. standard vs. mandatory flags, or a specific custom flagset. This would also allow us to run the debugger against the script_tests.json.