Process iterator in RPC result

[shargon]

Alternative to #577

[erikzhang]

If you return the values of the iterator directly, the developer will not be able to distinguish whether the method returns an array or an iterator.

[shargon]

If you return the values of the iterator directly, the developer will not be able to distinguish whether the method returns an array or an iterator.

I think that api users doesn't care about the low level type, they want the value. Otherwise you will need to know first that it's an iterator result in order to make a different call.

[erikzhang]

At least we should not modify its type.

{
  "type": "InteropInterface",
  "value": []
}

[satoshichou]

I tried modules, making everything to config is not a good way, and server can't modify config when it's running a node. I find more and more paramaters are added to config.json, it's too complex for reading and understanding for a new developer. I like the style of #577 much more. Should give rights to user but not limit it from server, that's why choose iteratror but not list, right?

[superboyiii]

As @satoshichou said, I totally agree. Don't make config too heavy and user could choose how much times iterator could run.

[shargon]

Don't make config too heavy

Anyway an output limit it's a good choice.

and user could choose how much times iterator could run.

The cost of this "feature" it's to choose a different method if you call an specific one, harder for developers, there are any use case for return an specific index? the server will process all of them in order to fetch them, this process can be delegated to the client.

[steven1227]

I agree with @superboyiii and @satoshichou and I prefer that Pull Request

[erikzhang]

I support @shargon's solution. Because if you use index, it may cause data inconsistency. For example, if I call invokeiterator twice, the first request is 1~100, and the second request is 101~200. But if a new database insertion or deletion occurs between the two requests, the data returned between these two requests is inconsistent.

[nicolegys]

I'm confused about MaxResultItems. My test shows that when I set the value 10, if the amount of return result is more than 10, I will get this returned "stack": "error: invalid operation". I think it should return the first 10 result and give a message, to tell the user that there are still some data not showed.

[nicolegys]

There's a boundary case. If the value of MaxResultItems is just enough to iterate over the data, the return result still shows the warning.

[shargon]

Now is not a MaxResultItems it's a MaxIteratorResultItems

[erikzhang]

Yes. We should rename it.

[nicolegys]

Test pass, just need to rename the parameter.

[nicolegys]

Since this solution not supports index, when the amount of data is very large, does user need to run his own node to get the complete data?

[erikzhang]

Since this solution not supports index, when the amount of data is very large, does user need to run his own node to get the complete data?

Yes. I think public nodes are not suitable for running high-load tasks.