Skip to content

Import / Merge WebGenai

💡 TL;DR - Import WebGenAI Project

You can import WebGenAI projects, merging their data models and rules into a dev project.

als genai-utils --import-genai --using=<WebGenAI-Project>

Through experience, you will discover that the IDE maintains full logic automation, albeit using IDE completion instead of Natural Language. That raises the question: how do I balance work in WebGenAI and the IDE?

You will find WebGenAI remains useful, particularly for updating the data model (e.g., as required for logic). The system provides services described here to sync WebGenAI changes with IDE changes.

If you elect to focus on IDE development, analogous services are provided with Rebuild from Model / Database, and Alembic support.

 

Coordinating Parallel Dev Streams (Multi-Team Development)

A classic challenge: enable ongoing parallel development with both the LOB and Dev teams. It's enabled by declarative technology, where the integration is done with software, not manual effort.

It works like this:

  1. The project begins with the Kickstart: several iterations to get solid requirements.

    • Uses Natural Language and Declarative Rules

  2. The project is exported: the Dev Team begin work on the Custom UI, Enterprise Integration, etc

    • This is not the end of declarative: logic is either or both of Natural Language, and Python DSL

  3. As depicted in the divergent arcs: there are now 2 versions of the project. Both can, in parallel, introduce new rules and attributes. This is not just "Team Development", it is TeamS Development.

    • LOB Parallel: (WG_Team): new rules & attributes
      • Observe: WG is not a day-1-only pilot...
      • They continue to use the Natural Language Web-Based interface (WebGenai)
    • Dev_Team: their own new rules & attributes, using either genai Natural Language and/or alembic...
      • In addition, the server team also uses Python (and) and libs as required, e.g. for enterprise integration.
        • They do this in their favorite IDE, with tools such as GitHub etc.
        • Observe the logic remains at a high lev
      • The UI team can use familiar tools for Custom UIs. These
        • leverage the API (ready day 1, so they are not blocked on API Dev), and
        • are dramatically simplified by automated backend rule-based logic

  4. Export 2: illustrates that the LOB WG_Team can export their project. The dev team can import it using the API Logic Server CLI.

    • This automatically integrates rules and attributes from both projects, updating the dev project with a new database and models.
      • This is virtually impossible with procedural code, because developers must manually assess the execution dependencies and order the logic properly. It time-consuming, complex, and error-prone - just like post deployment maintenance.
      • The integration is automatic and "safe" because logic is expressed in a declarative rules/models for which ordering is automatic.
    • It also rebuilds the test data, per your rules (e.g. sum/count values)
    • The process supports multiple exports.


genai-process


Exploring Import


Setup: Manager pre-installs Import Sample

When you create the manager (strongly recommended), the system installs 3 sample projects you can use to explore import.

  1. Base Project is GenAI_no_logic. No rule-based attributes. See system/genai/examples/genai_demo/wg_dev_merge/base_genai_demo_no_logic. It's not really used, just provided as a reference.

  2. Dev Project was created with export-1, and has added rules for carbon_neutral. It is ready for export-2. See system/genai/examples/genai_demo/wg_dev_merge/dev_demo_no_logic_fixed

  3. WG project has continued from export-1 to add our standard customer.balance rules. It is ready for export-2. See system/genai/examples/genai_demo/wg_dev_merge/wg_demo_no_logic_fixed.

    • It has an docs/export/export.json, which describes the data model and rules from the WG project. This is used for import.

The naming convention is that these started with no rules, had rules added, and were "fixed" by Genai-Logic to update the data model.


Usage

Imports are performed from with the dev project, using the import-genai CLI command:

cd system/genai/examples/genai_demo/wg_dev_merge/dev_demo_no_logic_fixed
als genai-utils --import-genai --using=../wg_demo_no_logic_fixed
Observe: 1. The data model (system/genai/examples/genai_demo/wg_dev_merge/dev_demo_no_logic_fixed/database) contains Customer.balance and Product.carbon_neutral 2. The test data has been updated to include these attributes, with proper values

In this example, als genai-utils --import-genai ... will leave things in this state:

genai-process

The import-genai command creates the docs/import directory and the following files, as shown above:

  • request.json is sent to ChatGPT. It contains both models, and a command to merge them
  • response.json is the merged model. It should reflect the attributes from both sides, as shown
  • The response is translated to system/genai/examples/genai_demo/wg_dev_merge/dev_demo_no_logic_fixed/docs/import/create_db_models.py
  • The system creates docs/import/create_db_models.py/create_db_models.sqlite by executing the file above.
  • The system then uses this to update the dev project:
    • update the dev database/db.sqlite and
    • Runs --rebuild-from-database. This updates the model, the api, etc from the new database.
    • It's good practice to verify these. Make sure all the attributes from both sources are reflected in the updated database and models noted above.


Restart option for failure recovery

It may fail, requiring either a re-run or an import-resume:

  • Re-run is indicated if the data model is missing attributes, incorrect or imcomplete.

    1. make sure to get initial system/genai/examples/genai_demo/wg_dev_merge/dev_demo_no_logic_fixed/database/models.py (eg, update from models_for_resume.py)
    2. delete or rename the docs/import directory.

  • import-resume can be used if you can repair the file below, e.g., a minor syntax error.

    1. fix system/genai/examples/genai_demo/wg_dev_merge/dev_demo_no_logic_fixed/docs/import/create_db_models.py
      • Note: you can run this standalone with your IDE to verify it. It should create create_db_models.sqlite in your docs/import directory.
    2. make sure to get initial system/genai/examples/genai_demo/wg_dev_merge/dev_demo_no_logic_fixed/database/models.py (eg, update from models_for_resume.py)
cd system/genai/examples/genai_demo/wg_dev_merge/dev_demo_no_logic_fixed
als genai-utils --import-genai --using=../wg_demo_no_logic_fixed --import-resume


Appendices


Ground Rules

  • No Dev_Team -> WG_Team integration (just deploy Dev_Team version, and use)
    • Dev team code cannot be integrated into WG - dependencies, libs, integration, ...
  • WG_Team - serial dev (as now)
  • WG_Team logic files are separate from Dev_Team (eg, using logic/discovery)
  • sqlite only, for now (presume upgrade to 'some other db' is doable later)
    • Tyler, what were the issues you mentioned in sqlite that forced you to use PG?
  • All Dev_Team and logic generations are finished before merge-G