It's just not possible to use maps in invokers/actors without this. And maps
have too many combinations to try pushing them into a type switch, that's
where reflection kicks in and solves it easily.
Signed-off-by: Roman Khimov <roman@nspcc.ru>
Pros:
* less code
* handles more types
Cons:
* slow
This code is not likely to be on the hot path and it is exactly the one used
by actors for making calls of various kinds. Supporting more types is more
important here than raw speed.
Signed-off-by: Roman Khimov <roman@nspcc.ru>
It's _very_ annoying to not be able to use []string types properly:
test invocation failed: unsupported parameter []string
But the same thing can happen to any other slice, so accept slices of every
basic type we accept above. Reorder tests to match implementation switch as
well.
Signed-off-by: Roman Khimov <roman@nspcc.ru>
Everywhere including examples, external interop APIs, bindings generators
code and in other valuable places. A couple of `interface{}` usages are
intentionally left in the CHANGELOG.md, documentation and tests.
Share parameters parsing code between 'contract invokefunction' and
'vm run' commands. It allows VM CLI to parse more complicated parameter
types including arrays and file-backed bytestrings.
Value of PublicKey parameter always stores public key bytes, not the
deserialized representation. All other code (CLI parameters parsing with
its NewParameterFromString, Parameter unmarshaller, etc.) is based on
the idea that value of PublicKey is []byte.
Which is almost like a NeoFS's toStackParameter() on steroids (except it
doesn't mess with noderoles package, it can be casted to int). RPC client's
Invoke* functions expect Parameters, so make it easy to create them.
They were first introduced in a058598ecc and
then carefully moved in 648e0bb242, but it looks
like they were never used by any external code. This code can be useful on the
server, but the server has its own params package to deal with
parameters. Clients usually create Parameters and then get results as
stackitem.Items, so they don't use this code either. So there is zero point in
keeping it.
It's not needed any more with Go 1.13 as we have wrapping/unwrapping in base
packages. All errors.Wrap calls are replaced with fmt.Errorf, some strings are
improved along the way.
MarshalJSON should be defined on structure (not pointer), as we use
structures to marshal parameters (e.g. in NotificationEvent and
Invoke of RPC result package) and never use pointers for that purpose.
Also added marshalling of nil array into `[]` instead of `null` to
follow C# implementation.
Neo 3 can emit Null in its transfer notifications in `from` or `to` fields
when minting/burning tokens (unlike Neo 2 that emitted util.Uint256{} for this
case), then it gets converted to Parameter as AnyType and we have to JSONize
it somehow for proper RPC functioning.
We actually have to do that in order to answer getapplicationlog requests for
transactions that leave some interop items on the stack. It follows the same
logic our binary serializer/deserializes does leaving the type and stripping
the value (whatever that is).
There's a bug after #785: smartcontract.Parameter of type hash160 should
be marshalled in LE (as default marshaller for uint160 does) instead of
BE, so fixed.
Fixes#809.
Basically, there are three alternative approaches to fixing it:
* allowing both []byte and string for ByteArrayType value
minimal invasion into existing code, but ugly as hell and will probably
backfire at some point
* storing string values in ByteArrayType
incurs quite a number of type conversions (and associated data copying),
though note that these values are not changed usually, so dynamic
properties of []byte are almost irrelevant here
* storing only []byte values in ByteArrayType
makes it impossible to use them as map keys which can be solved in several
ways:
- via an interface (Marshalable)
which is good, but makes testing and comparing values in general harder,
because of keys mismatch
- using serialized Parameter as a key (in a string)
which will need some additional marshaling/unmarshaling
- converting MapType from map to a slice of key-value pairs
not a bad idea as we don't use this map as a map really, the type
itself is all about input/output for real VM types and this approach is
also a bit closer to JSON representation of the Map
