Redesign Python 3rdparty dependency management (lockfiles)

[Eric-Arellano]

A catch all for the projects described at https://docs.google.com/document/d/1bCYb0UQZx9a-9tAagydCN_z3826QRvz_3aVnXKSNTJw/edit and tracked by #11898 and #11165.

As the initial prototypes get more fleshed out, the TODOs linking to this ticket should likely be split up into more granular tasks.

[Eric-Arellano]

Some updates on this umbrella ticket for tool lockfiles (#11898) + multiple user lockfiles (#11165).

Tool lockfiles have landed and are stable in 2.7! Only three remaining tasks:

• Add tests for setup_py_commands field #12658 (0.5 days)
• Add tests for repl.py #12657 (0.5-1 days)
• Document and blog (1-2 days)

Multiple user lockfiles now have a proof of concept at #12734. Some follow-ups are tracked in https://github.com/pantsbuild/pants/projects/20. Three main streams of work there, which I estimate as:

• 4-6 days for finishing implementation of multiple user lockfiles, including platform handling and compatible_resolves
• 4-5 days for lockfile invalidation work Chris is doing
• 2-3 days for documentation and blogging

In total, that is about 3-4 person weeks. @chrisjrn and I will continue working on this project as our top priority.

I am not including Pex work in the estimate (pex-tool/pex#1400), although @jsirois has made some exciting progress including pex-tool/pex#1437. I expect the Pex->Pants adoption to involve little Pants work because the generation rule is well encapsulated. But it will probably uncover some edge cases that require more Pex work.

[stuhood]

This might be completed...?

[Eric-Arellano]

Yep!