TM1Doc

UNSTABLE! use @ your own risk

Requirements

docker docker-compose

Usage

Move config.example.py to config.py and edit to match your environment.
run docker-compose up to start

Assertions

• words in Cube names start with are capital letter or can be capitalized
• you have enabled TM1s REST api (HTTPPortNumber=xxxx),
  • make sure you have configured your firewall to allow xxxx
• you have a user that is allowed to read all relevant objects on the tm1 instance
• you don't dynamically write to multiple cubes from the same process. eg. you don't do "CellputN(vCubeName, ..." where vCubename is NOT a constant.
• you assign cube names to constants in the prolog of load processes

Licence

It is MIT so steal it, tune it, break it and make it better.

Feature Requests

• Make sure cache clearing actually works ...
• Migrate ALL THE THINGS to tm1py

• Stop docker from writing shit as root into local folder

• fix name of flatfile objects - currently backslashes from windows fs don't show

• make dimensions a valid target for processes to write to (parse for dimensionelementinsert, dimensionelementcomponentadd, attrputs, attrputn)

• create sample instance

  • steal and ammend example from big blue? Which one is good?
  • run automated integration tests on this instance

• improve test coverage (up from 0)

• use some odata library to query tm1

• create objects smarter via some json magic

• comply with Pythons PEP8 (formatting) and PEP257 (docstring)

• Rank time complexity for subgraphs

• create multiple views on the output (via python cmd interface ?)

  • Create output for D3.js (example: http://bl.ocks.org/mbostock/4339083) [Edward Stuart]

• get object sizes (including feeders ...)

  • on disk
  • in ram

• Scheduling and parallel execution of processes

  • via "packages" as independent components of the dataflow graph
  • via knapsack-like optimizer
    • general case is NP hard
    • in this case greedy should work brilliantly
  • add estimated runtime
  • maybe create input file for cubewises hustle (https://hustle.codeplex.com/) needs c#/.net :(

• logfile parser

  • integrate into class objects?

• Add Source Code Control system via git & REST

  • also use it as a form of CI
  • maybe simple cmd line script to push process from filesystem to server via rest api?

• perform various model checks

  • unused
    • dimensions
    • cubes
    • processes
    • groups
  • high feeder:data ratio
  • automated "overfeeding"-cubes (http://www.ibm.com/developerworks/data/library/cognos/financial_management/analytics/page620.html #5)
  • cube and dimension filesizes
  • dimension: one element consolidations
  • dimension: hierarchy checks
  • dimension: elements without data and rule calculations (feeder flags)
    • unused element ratio = elements with data / all elements
  • cube: dimension order and sparsity checks
  • load time analysis
    • cubes
    • feeders (rules)
  • empty subsets

• create possibility for processes to have multiple targets

• get ignore_tech_objecs to work

• make the output graph pretty (dimensions clutter it up)

• Check whether MultiDIGraph makes sense -> nope!

• Dimension and Dimension -> Cube mapping

• REST Connection

Usefull Links

• description of REST API: http://swagger-ui.cubeac.com/#/
• alternative, easy tool written in PERL by some smart person: http://www.bihints.com/graphing_tm1_data_flow
• usefull regex playground: https://regex101.com/