<@U06A03HV1> "named interpreter constraints" a la ...
# developmenth
hundreds-father-404
08/25/2021, 2:59 PM@witty-crayon-22786 "named interpreter constraints" a la your "named resolves" idea is a great way forward to address my concerns about enabling
interpreter_constraints[python-setup]interpreter_constraints="py2_and_py3"👍 2
hundreds-father-404
08/25/2021, 3:01 PMIt addresses my two concerns!
1. It's no longer as easy to unintentionally use an advanced feature. You can only use pre-blessed ICs, and if you add a new one to pants.toml hopefully the right people will see it.
2. We can quickly determine if the repo is using >1 IC by looking at the option, rather than scanning the whole repo
hundreds-father-404
08/25/2021, 3:06 PMCc @early-jelly-55072 @helpful-lunch-92084, any thoughts?
e
early-jelly-55072
08/25/2021, 7:07 PMNate solved this using macros but the only setting was ICs. How would this improve on just using macros?
h
hundreds-father-404
08/25/2021, 7:59 PMMacros are valid too, but this is more first-class. It does the right thing by default, rather than requiring that you realize you should use a macro and then requiring that all your developers are disciplined in using that macro
(also unlocks some performance improvements in codebases using only a single interpreter constraint for whole project)
👍🏽 1
hundreds-father-404
08/25/2021, 8:23 PM3 Views