specs: VEP-1739 Support for Multiple Python Versions

[doks5]

This change introduces VDP-1739, which aims at proposing an improvement to Versatile Data Kit by adding support for using different python versions for data job deployments.

Testing Done: N/A

[antoniivanov]

Thanks for suggesting and driving this change. The proposal is generally well-structured and addresses the problem at hand effectively.

I have made some suggestions to further improve the proposal.

[ivakoleva]

LGTM! Thanks for driving this effort for improving usability in terms of granular data job runtime environment configurations.

[antoniivanov]

Looks good to me. I made suggestion to improve sequence diagram only.

[antoniivanov]

I approved it because I think the Motivation, Requirements and High level design sections are good to be merged.

For the detailed design I think we should cover

• Troubleshooting - what are the failure modes and what we can do to resolve them, what users can to do
• Test Plan
• Upgrade / Downgrade Strategy
• Database data model changes (we need to keep python version somewhere).

It's fine if some details are not yet clear and can be filled in the middle of the implementation when they become clearer

[doks5]

I approved it because I think the Motivation, Requirements and High level design sections are good to be merged.

For the detailed design I think we should cover

* Troubleshooting - what are the failure modes and what we can do to resolve them, what users can to do

* Test Plan

* Upgrade / Downgrade Strategy

* Database data model changes (we need to keep python version somewhere).

It's fine if some details are not yet clear and can be filled in the middle of the implementation when they become clearer

Hi, @tozka

Indeed, more details will be added with subsequent PR. For this one, I've updated the diagrams and added the initial parts of the detailed design and the public API changes.

[mivanov1988]

Few comments regarding the sequence diagram:

• The CS first starts the deployment and then returns the status to the client that the deployment is started.
• What actually do the following operations - "Request data job base image with specified python version" and "Request data vdk image with specified python version"?
• Registry -> Docker Registry

[doks5]

Few comments regarding the sequence diagram:

* The CS first starts the deployment and then returns the status to the client that the deployment is started.

* What actually do the following operations - "Request data job base image with specified python version"  and  "Request data vdk image with specified python version"?

* Registry -> Docker Registry

Hi, @mivanov1988

Just updated the diagram to make the actual flow clearer.