[MAINT] Change coverage uploader installation source
Codecov Report
Merging #861 (f0682dd) into main (ee75517) will increase coverage by
6.98%. The diff coverage isn/a.
:exclamation: Current head f0682dd differs from pull request most recent head 1df5d35. Consider uploading reports for the commit 1df5d35 to get more accurate results
@@ Coverage Diff @@
## main #861 +/- ##
==========================================
+ Coverage 30.01% 36.99% +6.98%
==========================================
Files 452 196 -256
Lines 39269 11584 -27685
==========================================
- Hits 11786 4286 -7500
+ Misses 27483 7298 -20185
| Impacted Files | Coverage Ξ | |
|---|---|---|
| libraries/fiff/fiff_info.h | 0.00% <0.00%> (-100.00%) |
:arrow_down: |
| libraries/fiff/fiff_stream.h | 0.00% <0.00%> (-100.00%) |
:arrow_down: |
| libraries/fiff/fiff_info_base.h | 0.00% <0.00%> (-100.00%) |
:arrow_down: |
| libraries/fiff/fiff_named_matrix.h | 0.00% <0.00%> (-100.00%) |
:arrow_down: |
| libraries/connectivity/connectivitysettings.h | 0.00% <0.00%> (-100.00%) |
:arrow_down: |
| libraries/inverse/hpiFit/hpifit.h | 0.00% <0.00%> (-50.00%) |
:arrow_down: |
| libraries/connectivity/network/network.h | 0.00% <0.00%> (-50.00%) |
:arrow_down: |
| libraries/fiff/fiff_evoked.h | 0.00% <0.00%> (-14.29%) |
:arrow_down: |
| libraries/fs/surface.h | 0.00% <0.00%> (ΓΈ) |
|
| libraries/fs/annotation.h | 0.00% <0.00%> (ΓΈ) |
|
| ... and 272 more |
I really like this @LorenzE , but in an effort for openness, just to let you know, there is a github action wrapping this bash script you're downloading. Let's see if this is clear: There is a python based uploader which will be deprecated in feb 1, 2022. There is a bash uploader which in its version 1 will also be deprecated in feb1, 2022. But there is another bash uploader (a v2 of the previous one) which is the "thing to use". There is a github action that uses this bash uploader, although wrapped in javascript with all its glory.
I'm all in to use this. But there is a github action related wrap ππ» ππ» https://github.com/codecov/codecov-action
Now that you know about this other option..... I like better the bash script you propose. π It would be great to know the actual coverage locally, is there a way to do that with this bash? have you done ./codecov --help?
I use the uploader from the official documentation. I guess this is the correct one. I am aware of the github action. The action might be tricky to use in our scenario since we upload the codecov report (via the uploader) after every executed test. codecov is able to merge these individual reports. I am not sure how we can upload/merge reports when using the action. I do not know if codecov can produce reports locally. afaik this is not possible. We would need to use a different reporting tool for that I guess.
Yes, that is the correct one, as mentioned in the documentation you link and the one i link too. Both from codecov.
We can change our script to run codecov only once, very easily. If you think this helps, it should be easy. Do let me know.
When you --help the codecov uploader you get a bunch of options. it seems like there is a dryRun version that lets you run codecov but doesn't upload the reports. But as long as i've been able to test it, it doesn't print out the final report or nothing similar.
So I think this helps, but I'm quite dissatisfied with the 'distance' between the testing environment and the developer during it's everyday developing effort / tinkering with the code. You can't make changes, and then wait for minutes (at least) to see their effect in terms of coverage. In this sense, this is not what we need.
But it is clearly something we need, at least to get through the feb 1st, due date. So thank you.
Let me share what I like:
- web based.
- automatic integration into github.
- I see. I like the effort.
We do not want to let codecov run only once but instead after every test run.
Maybe we can introduce a new tool for local reporting or change to a completely different tool at all. I think codecov has the nicest github integration though.
We do not want to let codecov run only once but instead after every test run.
Why? each test generates its own report file that is later bundled together with whatever tool. Codecov seems to upload each file and do the bundling backend. That would explain why we can't run it locally.
Maybe we can introduce a new tool for local reporting or change to a completely different tool at all. I think codecov has the nicest github integration though.
And web integration too. I also like it. But I wonder what other options we have.
These changes are not final yet. Please do not merge until this PR looses the draft tag.
Added to #871 .