Closes test related issues

[kyrcha]

Closes:

• Improve oracle update #51
• DB tests fail on master #47
• Add db default for tests #46
• Add the Windows build environment to Travis builds #25

I have also added a new "constructor" in github/downloader.go because the storer is not exposed and I could not get access to the storer for counting the various objects in the recording script.

---

• I have updated the CHANGELOG file according to the conventions in keepachangelog.com
• This PR contains changes that do not require a mention in the CHANGELOG file

[kyrcha]

From https://docs.travis-ci.com/user/best-practices-security#steps-travis-ci-takes-to-secure-your-data:

The beta Windows support does not obfuscate secure environment variables leaked into the build log. Please keep reading the next section, on how to avoid leaking secrets to build logs

So I have removed the windows env for possible leaks. Will reintegrate when out of beta.

[smacker]

It's better to create small PRs for each issues than 1 big pr that does everything. That way we can merge fixes and improvements faster.

In general LGTM but I think we need to change constructors to make it accept store.

[kyrcha]

OK will make smaller contributions next time.

In general LGTM but I think we need to change constructors to make it accept store.

OK so I am waiting for this change?

[kyrcha]

I think I did something wrong again. I'll try to fix it.

[smacker]

codecov: -64.22%

OK so I am waiting for this change

nah. it's just a recommendation.

also please use rebase.

[se7entyse7en]

👍 for smaller PRs and rebase.

[kyrcha]

Please do not merge yet. Looking into the low codecov issue

[se7entyse7en]

I only don't know if testutils.Tests is a good name as they're actually not tests themselves. Maybe something like testutils.TestsOracle? But other than that LGTM.

[kyrcha]

Actually I've pinpointed the problem in this part so I will refactor it and update.

[kyrcha]

OK, I think this is ready. The lower coverage comes from extra code added in between. Will rise after closing issue #40. From the next PR, I will work on the git workflow discussed.

[se7entyse7en]

I still don't understand the coverage thing. Shouldn't it increase? Is it caused only by having added the NewMemoryDownloader function?

[kyrcha]

retry.go has 0% coverage on codcov, while locally it has 40%+. I will further investigate.

[smacker]

@kyrcha do you run locally with a token? Because I see this PR is created from you fork. So maybe it executes only offline tests and retry is covered only in online tests?

[kyrcha]

Yes I run it locally with a token. OK I see your point.

[smacker]

Yes I run it locally with a token. OK I see your point.

most probably we need to open the PR from a branch in metadata-retrieval or just ignore this codecov report and merge as is (as we know it didn't really decrease)

[kyrcha]

OK I am closing this PR and re-submitting it from a branch.

[se7entyse7en]

Oh I didn't realize this was from your fork, I understand now 👍