add tests for releasesync

[ncabatoff]

Here are some tests for releasesync to address #1037. Beyond the additional tests, I:

• killed off some unreferenced symbols, made some helper methods private
• defined the interface Releaser, made ReleaseChangeSync depend on it
• expanded godoc comments
• added some calls to errors.Wrap
• added a dependency on go-cmp

Questions:

• Is the ignoring of errors in calls to addDeletedReleasesToSync/addExistingReleasesToSync intentional?
• Why do we do the 'git pull' after having computed what work to do, rather than before?

I'd be happy to continue down this path and add more tests if my approach is satisfactory to the maintainers, though I think sync() is the only other method in this package that's particularly wanting testing.

[squaremo]

LGTM. I admit I have not tried to fully understand the code as it stands. But some test coverage is very welcome!

[squaremo]

You'll need to rebase on master and rerun dep, since there was another PR that modified Gopkg.toml.

You may as well squash a couple of the fixup commits while you're there ;-)

[ncabatoff]

Ok, will rebase/dep/squash as you say. I had a couple of questions (re potential bugs) about the code in my PR; should I open separate issues for them, or ask on slack?

[squaremo]

Is the ignoring of errors in calls to addDeletedReleasesToSync/addExistingReleasesToSync intentional?

Probably not; I think it's reasonable to leave a FIXME comment there and move on for now.

Why do we do the 'git pull' after having computed what work to do, rather than before?

Yes that is odd.

Forensics (git blame) indicates that it was part of a fairly large set of changes regarding how git repos were used. Given that, and the alpha status of the helm operator, it's plausible that this particular code is logically wrong but (mostly) works. I reckon this one is worth an issue.

[ncabatoff]

Ok, added the fixme comment. I did the squash/etc, but please have another look because I'm pretty new to multi-user git workflows and it's possible I messed something up. I'll open an issue for the late pull.

[ncabatoff]

#1092 for the odd pull timing.

[squaremo]

Brill 🌟 thanks @ncabatoff