remove upper bound for multidict

[all2ham]

Is there a reason this had a hard pin? It would be beneficial to set this range as wide as possible to avoid conflicts with Sanic users.

[codecov]

Codecov Report

Merging #1969 (d0f0e73) into master (b4fe2c8) will decrease coverage by 0.18%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #1969      +/-   ##
==========================================
- Coverage   92.43%   92.24%   -0.19%     
==========================================
  Files          29       29              
  Lines        3263     3263              
  Branches      574      574              
==========================================
- Hits         3016     3010       -6     
- Misses        168      172       +4     
- Partials       79       81       +2     

Impacted Files	Coverage Δ
sanic/server.py	80.48% <0.00%> (-1.13%)	⬇️

---

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 b4fe2c8...d0f0e73. Read the comment docs.

[ahopkins]

As long as the tests pass, I have no problem with this.

[ashleysommer]

I know in the past we've had to pin Multidict to specific versions because major version changes brought in API-level incompatibilities (specifically in CIMultiDict, that we use for Headers).
Eg, Sanic 19.3.x, 19.6.x, and 19.9.x each had multidict>=4.0,<5.0 to avoid the incompatibilities introduced in 5.0.

For Sanic 19.12.x we adapted to the Multidict 5.0 changes, and updated the dependency, but pinned it to ==5.0.0 to avoid further changes. This is a side effect of the fact that most Python projects, even big popular ones, don't adhere to Semantic Versioning, at least not to the extent that it is used and enforced in the Javascript ecosystem. Having said that, most projects do limit their breaking changes to Major version increments.

Thankfully it looks like Multidict is stabilised enough now that changing the requirement to >=5.0.0 seems not too scary. Maybe we could make it >=5.0,<6.0, just in case?

[ashleysommer]

@ahopkins
There is a potential argument for backporting this change to 19.12LTS: #1977