CI: use conda incubator in benchmarks

[fangchenli]

The benchmark ci currently has the longest runtime. Using conda incubator would save about 6-8 min.

[jorisvandenbossche]

@fangchenli this looks like a nice clean-up in any case (thanks!), but how would it save 6-8 minutes? (it currently does not seem to take less time as on master, unless the caching is not yet working here in this PR?)

[fangchenli]

@fangchenli this looks like a nice clean-up in any case (thanks!), but how would it save 6-8 minutes? (it currently does not seem to take less time as on master, unless the caching is not yet working here in this PR?)

My mistake... It's more like 3 minutes. The caching is not working in this PR (maybe it doesn't work on PRs?). The two database builds are using conda incubator currently. It takes less time to setup the environment on those two builds. So I think caching is working there.

[jreback]

caching is only enabled in the master branch pushes

[jreback]

weirdly the doc build is failing

[fangchenli]

weirdly the doc build is failing

doc/make.py --warnings-are-errors | tee sphinx.log ; exit ${PIPESTATUS[0]}

PIPESTATUS[0] is 1 now. Not sure why. Looking at the output of doc/make.py --warnings-are-errors, nothing is failing.

update

I missed this error.

>>>-------------------------------------------------------------------------
Exception in /home/runner/work/pandas/pandas/doc/source/whatsnew/v0.8.0.rst at block ending on line 201
Specify :okexcept: as an option in the ipython:: block to suppress this message
---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
<ipython-input-2-65b822729fc4> in <module>
----> 1 plt.figure()

NameError: name 'plt' is not defined
<<<-------------------------------------------------------------------------

[jreback]

maybe the doc build is not picking up the correct file / deps as it is in master?

[fangchenli]

maybe the doc build is not picking up the correct file / deps as it is in master?

I fixed the error mentioned above. Right now the outputs of doc/make.py --warnings-are-errors of this PR is identical to master's. Only the exit codes are different.

[fangchenli]

I'll revert the doc build and open an issue for it.

[jreback]

thanks @fangchenli

[jorisvandenbossche]

@fangchenli do you remember why you needed to make this change?

[fangchenli]

It caused an error while using incubator. But I didn't get the incubator to work for the doc build. I probably should revert it as well.

>>>-------------------------------------------------------------------------
Exception in /home/runner/work/pandas/pandas/doc/source/whatsnew/v0.8.0.rst at block ending on line 201
Specify :okexcept: as an option in the ipython:: block to suppress this message
---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
<ipython-input-2-65b822729fc4> in <module>
----> 1 plt.figure()

NameError: name 'plt' is not defined
<<<-------------------------------------------------------------------------