remove overlapping slots from app.Sanic, fix broken slots inherit of HTTPResponse

[ariebovenberg]

Summary

There are some __slots__-related problems in sanic:

$ slotscheck -m sanic -v
ERROR: 'sanic.app:Sanic' defines overlapping slots.
       - name (sanic.base.root:BaseSanic)
ERROR: 'sanic.response:HTTPResponse' has slots but superclass does not.
       - sanic.response:BaseHTTPResponse

• Fixing the slot overlap saves a bit of wasted space
• Fixing the broken slots inheritance ensures HTTPResponse won't get a __dict__ and saves a bit of memory in the process.

Questions

regarding tests: probably not needed. I discovered the slots issue with slotscheck, a tool I maintain. Of course, If you like, I can add it to tox/CI as I've done for instagram/LibCST and dry-python/returns.

[prryplatypus]

Changes lgtm

[codecov]

Codecov Report

Merging #2387 (6f8581d) into main (d4fb44e) will increase coverage by 0.002%.
The diff coverage is 100.000%.

@@              Coverage Diff              @@
##              main     #2387       +/-   ##
=============================================
+ Coverage   87.107%   87.110%   +0.002%     
=============================================
  Files           60        60               
  Lines         5003      5004        +1     
  Branches       905       905               
=============================================
+ Hits          4358      4359        +1     
  Misses         472       472               
  Partials       173       173               

Impacted Files	Coverage Δ
sanic/app.py	89.668% <ø> (ø)
sanic/response.py	89.385% <100.000%> (+0.059%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d4fb44e...6f8581d. Read the comment docs.

[ahopkins]

Thanks. Can you add that to the GH actions pipeline?

[ariebovenberg]

@prryplatypus does the one test failure look at all familiar to you? In theory my change could have affected the cookies in HTTPResponse, but I doubt it would only show up on 3.7 in this way 🤔

[ahopkins]

Agreed. I've never seen that test be flappy, so it is suspicious, but I wouldn't worry about that yet. I'll take a look and see what's going on with it.

Thanks for your effort in reaching out and putting this together. We actually have a lot more slots usage coming around the corner in some refactors, and with all the nesting I appreciate your utility.

[ariebovenberg]

@ahopkins awesome! Let me know if you have any ideas for features/improvements.