remote: avoid expensive revlist operation when only deleting refs

[strib]

No description provided.

[codecov]

Codecov Report

Merging #567 into master will decrease coverage by 0.58%.
The diff coverage is 77.77%.

@@            Coverage Diff            @@
##           master    #567      +/-   ##
=========================================
- Coverage   77.98%   77.4%   -0.59%     
=========================================
  Files         129     129              
  Lines        9845    9851       +6     
=========================================
- Hits         7678    7625      -53     
- Misses       1328    1400      +72     
+ Partials      839     826      -13

Impacted Files	Coverage Δ
remote.go	74.04% <77.77%> (+0.4%)	⬆️
plumbing/transport/ssh/common.go	20.54% <0%> (-45.21%)	⬇️
plumbing/transport/ssh/auth_method.go	31.57% <0%> (-22.81%)	⬇️

---

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 bff1d06...467cb2a. Read the comment docs.

[mcuadros]

Any chance to cover this with a test?

[strib]

@mcuadros: any idea what kind of test I could write for this? My best idea would be to use mocks somehow to make sure that the revlist method is never called, but I don't see precedent for mocks in the test right now, and don't have time to build that infrastructure out myself.

I guess I could just test for correctness when all refspecs are deletes, rather than testing for performance, but that test would pass both with and without this change.

[strib]

I have no clue why this is reporting such a large test coverage difference. I'll try to throw together the simple correctness test and see if that helps.