penguinmenac3 / graphprogramming Goto Github PK

Execute the nodes in the correct order and trigger the next nodes when done. However, only trigger nodes which can actually be calculated.

Implement an input node that accesses the {arg} attribute in the command line parameter object and outputs it like an input node.

Implement a graph editor user interface.

A simple solution like Steve's Factory Manager does it would be totally ok. For a start.
http://www.minecraftforum.net/forums/mapping-and-modding/minecraft-mods/1293066-1-7-2-steves-factory-manager

For some weird reason the record node breaks the video input node. But only when the video input node is a camera (resource:0).

This bug does not occur when using files as video input.

Implement the principle that when there is no change in the data no calculation is required.

It is assumed that each algorithm is deterministic when given the same input, so it is stupid to calculate an algorithm twice on the same data.

However, this does not mean that algorithms must not be random but they should be deterministic or at least not break when not called multiple times with the same data.

This also gives a simple way to implement loops without explicitly adding them to graph programming.

Multi threadding gives the ability to wait for network inputs in a blocking mode on ticks.

Implement a node that illustrates how to use any existing executable no matter what language it is written in as a node.

Create subfolders for stdlib modules and allow prefixing them.
Remove the overhead Node ending names.

Basically order nodes into packages and make them typesafe, so that future editor versions can actually work with the type information. Some nodes may need splitting. (In some languages that means redundant code, however typesafety is nice to have wherever possible)

Packets would be (so far):
Default (graph, splitflow, any, ...)
Number (trigger, input, add, sub, equal, less, ...)
String (trigger, input, split, concat, equal, ...)
Json (parse, stringify)
Boolean (trigger, input, and, or, equal, not, ...)
Structures (for, while, if, leaky integrator)
Sys (print, network, executable, ..)
Arrays (trigger, input, concat, remove, foreach, get, insert, append, ...)
File (load, save)

Example:
InputNode -> Number.Input and String.Input
SplitNode -> String.Split

The webui executes graphs passed without checking if the provided graph name is a trap.

You can inject code to the commandline when passing a clever graph name.

A regex that allows subfolder/Graph or subfolder/subsubfolder/Graph2 is required.

Load and save graphs as json.

Implement a network node to show how to use network nodes.

A property view on the right side, when a node is tapped/clicked.
The property view should view properties of the nodes.

The editor should support subgraphs.

Using and editing subgraphs should be made far simpler.
Also reuse of subgraphs should be made simpler.

A node that calls another graph recursively.

Implement nodes for string operations.

Concat(a, b), IndexOf(str, ref, offset), Split(str, splitter), StartsWith(str, reference), EndsWith(str, reference), whatever seems usefull for a stdlib.

Correctly load the graph and the python implementations for algorithms.

Load the json containing a list with all availible nodes and their information.
Information means inputs and outputs (by name and type) as well as the code name of the node.

Look at the node information stored in the graphs. That information has to be loaded from some config file. Specification will soon be done.