Submodules

[jimhester]

This still needs some tests, but it works for simple cases, e.g. install_github("https://github.com/jonkeane/mocapGrip")

[codecov-io]

Codecov Report

Merging #138 into master will decrease coverage by 10.26%.
The diff coverage is 100%.

@@            Coverage Diff            @@
##           master   #138       +/-   ##
=========================================
- Coverage   93.26%    83%   -10.27%     
=========================================
  Files          27     28        +1     
  Lines        1514   1600       +86     
=========================================
- Hits         1412   1328       -84     
- Misses        102    272      +170

Impacted Files	Coverage Δ
R/install-remote.R	74.19% <100%> (-19.31%)	⬇️
R/install-github.R	64.82% <100%> (-33.15%)	⬇️
R/submodule.R	100% <100%> (ø)
R/install-git.R	20.73% <0%> (-71.96%)	⬇️
R/github.R	31.03% <0%> (-58.63%)	⬇️
R/deps.R	81.63% <0%> (-11.23%)	⬇️
R/utils.R	85.81% <0%> (-0.8%)	⬇️
... and 2 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6c7b63a...f296b2d. Read the comment docs.

[gaborcsardi]

@jimhester Is this ready?

[jimhester]

@gaborcsardi I was adding some tests, it is ready now if you could review it!

[jimhester]

Oh I guess the only other thing this could maybe have is a git2r version, right now it relies on having a command line git only. I can add it if you think we need it.

[gaborcsardi]

I think it is fine with system git.

[gaborcsardi]

Looks great!

[gaborcsardi]

I wonder if we should just include a collated rematch2 package, that would simplify things.

[gaborcsardi]

Need to skip if no system git. (If we are skipping the other tests that rely on system git.)

[jimhester]

Ok I added the re_match() from rematch2 (it is self contained in 1 file, so is easy to add), also skipped the test, thank you for the reminder. Let me know if there is anything else you think we should change.

[gaborcsardi]

I think it is good now.

I think if we want to be completely correct we would need to use the GIT_DIR (etc.) env vars. So maybe parsing the output of git submodule status etc. is a better approach? But we'll probably never run into this.

I don't think we need to change it now, just sg. to think about next time we write sg for git.

[jimhester]

The issue with that is the zipball we get from GitHub is not actually a git repository, e.g. it doesn't have a .git directory. That is why I can't just run git submodule init, git submodule update to retrieve the submodules and have to parse them by hand (believe me I tried this first :) ). So more complex things like GIT_DIR and defining submodules in a global config etc. actually don't matter for this case.

[gaborcsardi]

Got it. Yeah, these would only matter for local installs. All good then!

[jimhester]

Thanks for the review!