davidhstocker / graphyne Goto Github PK

This api command is used by Tioga and needs a testcase.

If the an entity exists in the catalog, but its meme is not derived from Memetic.DNA.Script or it has no valid script attached, then it currently throws a Exceptions.NoSuchEntityError exception; as if it were not there at all. This misinforms the user and developer as to the actual problem.

Graph.api.evaluateEntity() needs to throw a Exceptions.ScriptError exception in this case.

getCounterpartOverview() is a bit unwieldy. Let's shorten it before creating the first tag.

This api command is used in the condition and numeric python modules of Memetic, but there is no testcase.

Graphyne's internal map/reduce implementation needs to get testcases.

This is a transient, placeholder file. It has been many, many changelists (since long before Graphyne was forked as an independent project, from the Angela R<L interpreter project on sourceforge) since Graphyne created its validation results inside its own location. It now put them into ///Graphyne. The original purpose is to give a placeholder webpage to the designer while the repo was being validated.

1 - Unless the repo is massive, validation happens quickly.
2 - It is XHTML. How quaint!
3- Once I get around to making a tag and pypi setup, then this file will be generated within the python install path. This needs to be fixed BEFORE the first tag is made.

This means that you can't just start creating memes via the api and are forced to use a metameme. Metamemes should be optional.

Right now, it is very difficult to debug traverse queries. It requires the designer to run graphyne in debug mode and sift through the schema xml documents, to research the proper path and what would be expected. This is difficult, designer unfriendly and not possible with ad-hoc graphs. You can only do this if you defined the entire graph via schema and are not creating nodes via generic memes.

Graphyne needs a script command for returning the entities (and metadata) along a traverse path, along with their nearest neighbors.

This is a pure usability thing. It is just cleaner, shorter to write and Facade might wrongly give devs the idea that they have to use a factory pattern.

Add three new event types to State Event list. This will let us execute a script whenever a property changes, a link is added or eliminated.

The following three entres need to be added to State Event Restriction:
propertyChange
linkAdded
linkRemoved

The following api methods will also need to trigger these events:
addEntityLink()
removeEntityLink()
addEntityDecimalProperty()
addEntityIntegerProperty()
addEntityListProperty()
addEntityStringProperty()
addEntityBooleanProperty()

This API command needs a testcase

addEntityLink takes four parameters, two mandatory and two optional. entityUUID1 and entityUUID2 are mandatory. linkType is the third positional and the first optional. linkAttributes is the fourth positional and second optional.

linkType is a bit esoteric and its use case might not be a real world one. That is, I set up members of an entity in a schema with a subatomic link. They are visible when directly traversing from any of the entities created in this cluster, but the subatomic links are not traversed when travelling from the outside. They are effectively invisible to the rest of the graph.

This use case might be useful, sometime. linkAttributes are useful very often. It is poor API UX to have a (presumptively) commonly used optional parameter behind a rarely used one. The positions of these two need to be swapped, so that the order is:

entityUUID1, entityUUID2, linkAttributes, linkType

Hello, I am wanting to test graphyne with python 3.7, is it a problem with the python version or the windows?

from graphyne import Graph
Traceback (most recent call last):
File "", line 1, in
File "c:\users\bios-pc\appdata\local\programs\python\python37\lib\site-packages\graphyne\Graph.py", line 30, in
from .DatabaseDrivers import SQLDictionary
ModuleNotFoundError: No module named 'graphyne.DatabaseDrivers'

No module named 'graphyne.DatabaseDrivers'

Python 3.7.9 (tags/v3.7.9:13c94747c7, Aug 17 2020, 18:58:18) [MSC v.1900 64 bit (AMD64)]
Graphyne-1.2

This api method is supposed to get all of the nearest neighbors of a given entity. It uses a different code clock than getLinkCounterpartsByType(), but there is no testcase and in any cases where the smoketest does check for all nearest neighbors, getLinkCounterpartsByType(entityID, "*") is used. The latter is the canonical method, as using the single wildcard for meme type yields every nearest neighbor.

getLinkCounterparts() needs to become a wrapper for getLinkCounterpartsByType(entityID, "*")

getClusterMembers() returns a list of entities in the current entity's cluster. We need a method that returns the entity list and the link information, ideally in a format close t compatible with d3's Force Directed Graph ( d3.layout.force() ).

The following write log statements need test cases:
writeDebug()
writeLog()
writeError()

Memetic packages may contain sub-packages.

E.g. you could have a Bar subfolder inside a Foo repository folder and import only the Foo folder. A meme from another repo would access meme XXX inside of Boo via “Foo.Bar.XXX”. Right now, Graphyne supports this if the content contains only templates, but not if the subpackages contain Python code. This is because Graphyne walks the folder structure of declared repository folders and adds each to the Python path individually. Python3 does not allow us to import Foo and then import Bar.

We need to look into adding the repositories to the Python path and then using absolute import paths to clean this up.

Currently, in order to fully reset an entity to the original meme specification, you have to fire off two commands: , revertEntityPropertyValues and removeAllCustomPropertiesFromEntity. This is needlessly complex and we need a wrapper api method that fires off both.

Right now, link attributes are set when when a link is created between two entities. They are then frozen in time. This is not a critical showstopper... yet.

The Problems

1 - It could become a potential performance issue. Two entities may share n links. I can always add more links with new attributes. If I have a large number of link attributes, I may want to start condensing them where possible, in order to increase performance.

2 - Right now, updating a link is a two step process. use graph.api.removeEntityLink() to remove the existing link (filtering for the desired attribute) and then graph.api.addEntityLink(). What if the link has other attributes?

Gotchas and Things to Consider

We should not replace this workflow with another two step - get Link, update attribute - one. We're going to need an SQL style where clause; ideally using the same syntax that we already use with the link attribute filters of traverse paths.

Api command is used by Tioga.

API command is used by Tioga.

This API command is used by Tioga and needs a testcase

These two api commands need test cases. We'll need to be able to delete entities.

Currently, the State Event Script uses a numeric Memetic restriction, 0, 1 or 2. This is in no way human readable and presents a UX problem for the designer. Currently from setStateEventScript()

            if state == 0:
                self.initScript = function
            elif state == 1:    
                self.execScript = function
            elif state == 2:    
                self.terminateScript = function  

These integer values should be strings:
initialization
execution
termination

This should be done before the first tag is created, so that we don't introduce incompatible changes.

Right now, Graphyne is incapable of creating entities that are not already defined by a Memetic Meme. Memetic is awesome and all, but devs should be able to create a hello world graph without first writing XML, and presumably, they could get away with never, ever defining any schemas. We need to be able to define schemaless, generic entities at runtime, without a Memetic definition.

getAssemblyNetworkOverview is unwieldy and cryptic. Refactor it to getCluster(). This should be done before the first tag is created, so that we don't have any incompatible api changes.