Closed killy971 closed 9 years ago
The logic should be easy to implement in a similar to what was done to read and parse the page linked from coveralls response (See https://github.com/guillaume-nargeot/hpc-coveralls/issues/21).
I did a poor job of announcing this change. Thanks @guillaume-nargeot for recognizing this.
No problem Steve!
I've just fixed the problem, and will add retry logic later.
Feature branch: https://github.com/guillaume-nargeot/codecov-haskell/tree/fix-coverage-response-reading Fixed by 078bca0154f02222d0c1b70b6db7e2e422002f5f
Another change was required to fix the problem: using the CurlFollowLocation
option in order to follow the link returned by HTTP 301 responses.
This issue is now fixed as of 1c623d515fb1eb1873a3855b1b45f42899e8fb61
The following change was introduced on codecov server side, apparently by https://github.com/codecov/codecov-python/commit/2eab8f9a5b50dce28c6c187cf41961da5f4cda57: instead of containing the total coverage result as a value, the response to codecov initial post request now contains a url (
wait_url
element in the json content). Reading this url should return the coverage result in json format as in the example below:Here is the code showing how codecov-python implements the functionality to wait and read the total coverage result.