validate hinted attributes on subcontrollers

[jsouter]

Oversight from #187

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.33%. Comparing base (aecb4c3) to head (37da3e9).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #209      +/-   ##
==========================================
+ Coverage   89.31%   89.33%   +0.01%     
==========================================
  Files          45       45              
  Lines        2265     2269       +4     
==========================================
+ Hits         2023     2027       +4     
  Misses        242      242              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[jsouter]

We're currently not checking for non generic alias types e.g.

class MyController:
    my_attr: AttrR
    my_attr_unknown_access_mode: Attribute

Might also be good to validate hinted subcontrollers, though I guess they're not usually dot-accessed.

e: pushed something, setting as draft until I fix some logic with how hinting as an Attribute should work

[GDYendell]

I am just doing some updates to fastcs-eiger and it has some logic to check datatype and access_mode of hinted attributes that get introspected. I think these checks should be done in add_attribute (where currently it fails immediately because the hinted attribute already exists) and validate_hinted_attributes should only check for hinted attributes that weren't introspected. Then in fastcs-eiger we can just call add_attribute and fastcs will do all the checks.

What do you think? Possibly this could be a separate PR.

[jsouter]

I am just doing some updates to fastcs-eiger and it has some logic to check datatype and access_mode of hinted attributes that get introspected. I think these checks should be done in add_attribute and validate_hinted_attributes should only check for hinted attributes that weren't introspected.

What do you think?

Failing fast may be nicer, validate_hinted_attributes is currently only called on FastCS.serve. Would have to figure out a good way to track which attributes have already been checked and make sure we're not maintaining the validation logic in two places

[GDYendell]

That is true it might be more complicated. Let's get this PR merged and I will play around with the validation while I update fastcs-eiger.