vcsrepo module: is my patch faulty or are the test failures unjustified?

[Martin Heistermann]

Hi,

I have some vcsrepo resources that check out out a revision specified by SHA1 hash.

However, on every run 'git fetch' is run to verify that the correct revision is checked out.

I think that shouldn't be necessary, so I tried to patch vcsrepo to avoid such fetches in

that case:

https://github.com/mheistermann/puppetlabs-vcsrepo/commit/68cf48ec624dddfd1cdd034638131829fd2d46fb

After this change, several spec tests fail: https://gist.github.com/anonymous/11399436

You can find the source of the failing tests (current HEAD) here:

https://github.com/puppetlabs/puppetlabs-vcsrepo/blob/46588aa62d444c5f0a538cccec357a80cd191b93/spec/unit/puppet/provider/vcsrepo/git_spec.rb#L173

The test  "checking the revision property when its SHA1 is not different than the current

SHA should return the ref", which fits my setup, seems to fails because the changed

code does not do some git invocations required by the test, which i believe to be superfluous.

If I understand it correctly (I'm new to ruby and rspec), the other failing tests hand the string

'currentsha' to the tested code both as resource revision as well as 'git rev-parse HEAD'

output (L176, L178). I think that 'currentsha' would be the correct method return value, with no

git invocations except for 'git rev-parse HEAD'.

I am not sure if my code actually introduces wrong behavior (how?) or if the tests

failures are actually not justified and would be glad about your comments.

Thanks in advance!

Martin