Branch autogen entrypoint

[lightszero]

if you do not write a Main function in smart contract.
now we can add one automatic:

    class Contract_autoentrypoint : SmartContract.Framework.SmartContract
    {
        //there is no main here, it can be auto generation.
        //object Main(string name,object[])
        //{
        //    if(name=="call01")
        //    {
        //        return call01();
        //    }
        //    if(name=="call02")
        //    {
        //        call02(_params[0],params[1]);
        //        return 0; //null or "" or 0 or false , if call02 is void,will add return 0;
        //    }
        //    throw //if no method was found,make vm fault.
        //}
        public static byte[] call01()
        {
            var nb = new byte[] { 1, 2, 3, 4 };
            return nb;
        }
        public static void call02(string a,int b)
        {
            Neo.SmartContract.Framework.Services.Neo.Runtime.Log(a);
        }
    }

[igormcoelho]

This idea is really good. Congratulation Lights.

I reviewed most of this PR in the other PR... but I didn't like that one 😂 hahaha we need testing in both ways: standard and non-standard entrypoints.

[shargon]

Very nice work @lightszero ! i love it! my only concern is about verification, how we can deal with this, i think that it should return false in verification.

[shargon]

What about...

public static bool onVerification()
{
  // Logic here?
  return false;
}

[lightszero]

This idea is really good. Congratulation Lights.

I reviewed most of this PR in the other PR... but I didn't like that one 😂 hahaha we need testing in both ways: standard and non-standard entrypoints.

this idea is from eric.

[lightszero]

Very nice work @lightszero ! i love it! my only concern is about verification, how we can deal with this, i think that it should return false in verification.

err...,that is a question.

as eric said all neo3 contract is Main(string,object[])

but how about verification?
@erikzhang

[igormcoelho]

Congratulations for the idea @erikzhang 😂

if all is unified, fine.

I think your case is not a problem Shargon, because the method would push0 , and conversion already knows its a bool type, and ot would be returned normally. The only "edge" case is a void metgod, which is being treated as "return false", right lights?

This one is just when you dont declare your main, so it creates for you. The other PR is more problematic.

[lightszero]

I think we can merge this first.
and discuss about verification more.

maybe add some attribute to help compiler like

[Trigger(Verification=false,Application=true)]
void call01()
{
///
}

[shargon]

I was thinking exactly in this, an attribute!

[erikzhang]

For verification, you can write the Main() function manually.

[erikzhang]

We should also update the templates.

[shargon]

For verification, you can write the Main() function manually.

But by default, should return false

[igormcoelho]

I think that unused functions should FAULT on Application. If we have exception handling, this would help creating "inheritance". I don't know the plan yet, but pehaps an "special" FAULT state... like FAULT cause Exception="method not found".

[lightszero]

already fault，but fault is not have param in neovm. we can not set a fault reason.

[igormcoelho]

I think that FAULT reason is more important for a better Exception handling... in the future.

[igormcoelho]

Just to clarify, the problem is the THROW itself... sometimes you want to invoke something, and if it fails, you proceed with other alternative. So, we actually need exception handling. I'm just mentioning here because it would be useful together with this feature here... but perhaps it's time to start thinking on strategies to allow exception handling on vm.