[HOTFIX] Update sleep physionet to the complete dataset

[massich]

@hubertjb realized that in #5718 we used an incomplete version of the sleep dataset.
(It was not clear that this version was available so we also open MIT-LCP/physionet#97)

This PR updates the fetchers so that we use the full version of the data.

[agramfort]

pep8 ...

[agramfort]

@hubertjb I'll let you try first to confirm that it works for you.

thx @massich for taking a stab at this

[larsoner]

I am getting failures on master

https://travis-ci.org/mne-tools/mne-python/jobs/492796093

Is it related?

[codecov]

Codecov Report

Merging #5932 into master will decrease coverage by 37.35%.
The diff coverage is 89.47%.

@@             Coverage Diff             @@
##           master    #5932       +/-   ##
===========================================
- Coverage   88.72%   51.37%   -37.36%     
===========================================
  Files         401      401               
  Lines       72354    72351        -3     
  Branches    12154    12154               
===========================================
- Hits        64197    37169    -27028     
- Misses       5219    33069    +27850     
+ Partials     2938     2113      -825

[massich]

Is it related?

yes completely. They turned the data down. (see MIT-LCP/physionet#97 (comment))

regarding the fleck8 I was aware. We start migrating together with @hubertjb and we only changed BASE_URL for temazepam. We decided that we would continue tomorrow but I guess that now this is a hotfix due to errors in master.

I'm on it.

[larsoner]

Great, thanks. Will merge once Travis is happy again. If there is an example that uses this dataset, please (make sure it still works and) make some small change to it so that CircleCI will run it.

[massich]

There's a travis test that mimics what circle will find. Ah, ok now I see what you meant. modify the example to trigger the makefile.

[larsoner]

The mimic will usually not be as good/complete as having CircleCI actually download the data and run the example so we can examine the output. So please at least try running the example locally so that we don't break CircleCI with this PR

[massich]

If green this should go in as it is. There's a subsequent PR making use and testing the new data.

[massich]

The doc builds with no problem on my machine

[larsoner]

@massich can you fix the new flake errors? If not I can do it

[larsoner]

mne/datasets/sleep_physionet/age.py:17:1: E302 expected 2 blank lines, found 1
mne/datasets/sleep_physionet/_utils.py:149:5: F841 local variable 'subjects_url' is assigned to but never used
mne/datasets/sleep_physionet/tests/test_physionet.py:20:1: F401 'mne.datasets.sleep_physionet._utils.BASE_URL' imported but unused

[larsoner]

Thanks @massich

[massich]

I was AFK, Thx @larsoner for taking over

…

On Wed, Feb 13, 2019, 21:23 Eric Larson ***@***.***> wrote: Thanks @massich <https://github.com/massich> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#5932 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGt-49MbD1r6NPQLcybg0fYH91Pp-RVEks5vNHRSgaJpZM4a5g_h> .