Hi 👊

This is my first visit to this fine repo, but it seems you have been working hard to keep all dependencies updated so far.

Once you have closed this issue, I'll create seperate pull requests for every update as soon as I find one.

That's it for now!

Happy merging! 🤖

• new ignore param on all checkers
• new identifier class attributes on every plugin
• usage of self.logger in plugins
• new YAML format: identifier, name and description

• separation of privileges
• least privileges
• code clean

Currently it just checks for the object being a dictionary and having an 'analyzers' item. Must do more.

Such an attribute in checkers and providers could help combining and creating checkers and providers.
Indeed a checker expects some form of data and not another in order to work. Checkers and providers should be able to understand each other. This attribute should be an enforced minimal description of the format.

Currently, the user is forced to use the following names for groups:

• framework
• core_lib
• app_lib
• app_module
• broker
• data

It must be more flexible. It would allow a better integration in django-archan

I started asking a question, then mostly answered it. This is the condensed version number ~5. Essentially I just ran into a bunch of issues understanding how archan works and how to apply it to my project. I think my below notes might be helpful to others and could be optionally integrated to the README

Would you consider reordering the example commands? I think creating the CSV with dependenpy might be the most common use case and would be helpful if listed first. I wasn't sure where the CSV files were coming from until I dug further

# Generate CSV DSM data for archan (dependenpy must be installed)
dependenpy archan --format=csv archan_dsm.csv

# No configuration, read CSV data from file
archan --no-config --input archan_dsm.csv
# or read CSV data from STDIN
dependenpy archan --format=csv | archan --no-config

# Load configuration file and run archan (see Configuration section to know how archan finds the config file)
archan
# or specify the configuration file to load
archan --config my_config.yml

# Output the list of available plugins in the current environment
archan --list-plugins

I created a modified version of the archan config, but I had to comment out the names to prevent skipping the steps:

analysis:
  archan_pylint.PylintProvider:
    # name: Pylint
    arguments:
      pylint_args:
        - dash_charts
    checkers:
      archan.CodeClean:
        allow_failure: true
        arguments:
          threshold: 10
  dependenpy.InternalDependencies:
    # name: Software Architecture
    arguments:
      packages:
        - dash_charts
    checkers:
      - archan.CompleteMediation
      - archan.EconomyOfMechanism:
          allow_failure: true
          arguments:
            simplicity_factor: 2
      - archan.LayeredArchitecture
      - archan.LeastCommonMechanism:
          allow_failure: true
          arguments:
            independence_factor: 5
  Open Design:
    Source Code: true

Installed the dependencies and ran archan

> poetry add dependenpy git+https://github.com/pawamoy/archan-pylint.git git+https://github.com/pawamoy/archan.git -D

Updating dependencies
Resolving dependencies... (25.4s)

Writing lock file

Package operations: 8 installs, 0 updates, 0 removals

  • Installing dependenpy (3.3.0)
  • Installing lazy-object-proxy (1.4.3)
  • Installing wrapt (1.12.1)
  • Installing astroid (2.4.2)
  • Installing pylint (2.6.0)
  • Installing archan-pylint (0.1.0 ef4a609)
  • Installing tap.py (3.0)
  • Installing archan (0.3.0 9443064)

❯ poetry run archan -c .archan.yaml
TAP version 13
# TAP results for Pylint Provider: Issues per Module
ok 1 Code Clean (ALLOWED FAILURE)
# TAP results for Internal Dependencies
ok 2 Complete Mediation
ok 3 Economy of Mechanism
ok 4 Layered Architecture
ok 5 Least Common Mechanism (ALLOWED FAILURE)
# TAP results for Open Design
ok 6 Source Code

On a side-note, can you publish a new wheel to PyPi? The current one has a typo bug for .ingo and some API differences (notably analyzers vs. analysis)

stdin input is currently not detected. It is used if there is no arguments. Fix that.

What work is there to port this to windows?

• Layered architecture
• Code clean: some pyflakes and mccabe on app modules
• Open design: will be okay when all others will be implemented
• Least privileges: need user/object access matrix
• Separation of privileges: like above

Would it be too much to ask to include examples of CSV files to help people try to reverse engineer how this software is working please?

These tests should not be included in this autonomous package.

The DSM class should beneficiates from other methods related to adjacency matrices. See the wikipedia pages to list some functions that could be of interest, then implement them. We could have functions to change/sort the order of the elements, and also transitive closure function.