DOC: pin sphinx to 1.8.5

[jorisvandenbossche]

xref #26058

[jorisvandenbossche]

Context for doing this would be the styling changes introduced in sphinx 2, for which we should first update our custom css (described in #26058)

[jorisvandenbossche]

cc @datapythonista @TomAugspurger

If we would release right now, I would build the docs with sphinx 1.8.5 for the release.
Given that, the question is then: do we want to use 1.8.5 in the travis build as well, to make sure everything is working with that release (to match our release process), or do we rather already want to build with the newest possible sphinx to also catch possible new errors with newer sphinx? (both have value IMO)

[TomAugspurger]

I think pin to 1.8.5 for now.

[jreback]

Agreed pinning is fine (i think we have pinned sphinx in the past frequently)

[datapythonista]

+1 on pinning

I'd probably migrate to sphinx 2 when we change the theme. And surely after we fail the CI on doc warnings.

But can you also pin the version in environment.yml please, so we build the docs locally with the same version as travis.

[datapythonista]

and may be update the comment on pinning to why we're using 1.8.5, and duplicate it in environment.yml too (I'll propose getting rid of this deps file again soon).

[jorisvandenbossche]

Updated

[codecov]

Codecov Report

Merging #26781 into master will decrease coverage by <.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #26781      +/-   ##
==========================================
- Coverage   91.72%   91.72%   -0.01%     
==========================================
  Files         178      178              
  Lines       50779    50779              
==========================================
- Hits        46579    46575       -4     
- Misses       4200     4204       +4

Flag	Coverage Δ
#multiple	90.31% <ø> (ø)	⬆️
#single	41.21% <ø> (-0.06%)	⬇️

Impacted Files	Coverage Δ
pandas/io/gbq.py	78.94% <0%> (-10.53%)	⬇️
pandas/core/frame.py	96.88% <0%> (-0.12%)	⬇️

---

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 ea06f8d...124504f. Read the comment docs.

[codecov]

Codecov Report

Merging #26781 into master will increase coverage by 0.03%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #26781      +/-   ##
==========================================
+ Coverage   91.72%   91.76%   +0.03%     
==========================================
  Files         178      178              
  Lines       50779    50751      -28     
==========================================
- Hits        46579    46570       -9     
+ Misses       4200     4181      -19

Flag	Coverage Δ
#multiple	90.35% <ø> (+0.03%)	⬆️
#single	41.22% <ø> (-0.05%)	⬇️

Impacted Files	Coverage Δ
pandas/io/gbq.py	78.94% <0%> (-10.53%)	⬇️
pandas/core/frame.py	96.88% <0%> (-0.12%)	⬇️
pandas/plotting/_matplotlib/tools.py	78.4% <0%> (ø)	⬆️
pandas/core/generic.py	94.1% <0%> (ø)	⬆️
pandas/plotting/_matplotlib/style.py	85.96% <0%> (ø)	⬆️
pandas/plotting/_matplotlib/timeseries.py	66.32% <0%> (ø)	⬆️
pandas/core/internals/blocks.py	94.38% <0%> (+0.25%)	⬆️
pandas/plotting/_matplotlib/boxplot.py	75.59% <0%> (+0.48%)	⬆️
pandas/plotting/_matplotlib/misc.py	38.93% <0%> (+0.9%)	⬆️
pandas/core/internals/managers.py	95.21% <0%> (+1.34%)	⬆️

---

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 ea06f8d...d79c80d. Read the comment docs.

[datapythonista]

lgtm, thanks

[phamvantuong]

Ok travic job very good Vào 12 Th6, 2019 13:47, Marc Garcia <notifications@github.com> đã viết: @datapythonista commented on this pull request. +1 on pinning I'd probably migrate to sphinx 2 when we change the theme. And surely after we fail the CI on doc warnings. But can you also pin the version in environment.yml please, so we build the docs locally with the same version as travis. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#26781?email_source=notifications&email_token=AJLD4SJOYLFKZHYXYERY3PLP2CLWZA5CNFSM4HW5UGBKYY3PNVWWK3TUL52HS4DFWFIHK3DMKJSXC5LFON2FEZLWNFSXPKTDN5WW2ZLOORPWSZGOB3IOMOY#pullrequestreview-248571451>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AJLD4SJPJWRITYN4HBSMXULP2CLWZANCNFSM4HW5UGBA>.

[datapythonista]

In conda files the version is specified with just one equal (== is for pip, a bit confusing)

You'll also have to call ./scripts/generate_pip_deps_from_conda.py to update requirements_dev.txt, which is synched from environment.yml

[jorisvandenbossche]

Ah, thanks!

[datapythonista]

you also need to remove one equal here, other than that looks good

[jorisvandenbossche]

Just wondering: does it actually matter? (it was with double equals before as well)

[datapythonista]

not sure, but we've got a function to convert them from one to two when synching, if we can have two may be better to leave two everywhere and simplify the synching

[datapythonista]

thanks @jorisvandenbossche