added --dry-run option

[schnittchen]

This adds --dry-run support as requested in #1022.

I would need some help as to where to test this feature...

[coveralls]

Coverage increased (+0%) when pulling 9582260 on schnittchen:dry-run into a04fb17 on rspec:master.

[penelopezone]

This is very strongly opinionated so feel free to disagree with me better, but as conditionals get more complex I like turn them into ifs instead of unlesses. What do you think?

[schnittchen]

I am totally ok with changing this to "if". I realize that coming from a mathematics background, DeMorgan's rules are pretty hard-wired into my brain...

[schnittchen]

Or extract the condition out into a will_execute? method?

[JonRowe]

Normally I would agree with you @samphippen, especially if there's an else but I try to stick to the cleaner version and

unless pending or RSpec.configuration.dry_run

reads better than

if !(pending or RSpec.configuration.dry_run)
if !pending and !RSpec.configuration.dry_run

[JonRowe]

You can add specs to this in spec/rspec/core/configuration_options_spec.rb for checking the option is parsed correctly, and spec/rspec/core/example_spec.rb to check that an example group won't run if dry run is enabled.

You will also need to add a .feature to document the feature and will serve as an integration test to assert that everything works together.

[myronmarston]

I think this implementation will still run before(:all) and after(:all) hooks, which it shouldn't.

[schnittchen]

Correct. Argh. More research.

[schnittchen]

I think I got it. Should probably move tests around after I realized before/after(:all) block execution is a group thing. And please nitpick if you see things you don't like.

[JonRowe]

I think it would be better to assert on the output here, to give an example of what to expect, and I'm not sure we should be considering the tests as "passing" here. They're closer to being pending.

[schnittchen]

I would rather write something like "Then it looks like the tests are all passing" with the same assertion underneath.

[JonRowe]

The test's don't pass though, they don't even run, so to describe it as such is misleading in my opinion

[schnittchen]

I see. Would "Then it looks like the tests are all passing" solve that? Do you have a better suggestion?

[JonRowe]

Actually showing the output you expect, similar to how pending does it /features/pending/pending_examples.feature#L17

Edit (is what I suggest :) )

[schnittchen]

Updated.

[JonRowe]

You could probably collapse this, as none of the fails should ever be executed.

[schnittchen]

I don't know what you mean by collapse here.

[JonRowe]

A .feature file should describe the behaviour of the feature it's describing, having a large empty spec file is slightly unnecessary to that aim, you only really need to prove none of the hooks run and an example doesn't run. A more compact file would just have the hooks and a single example.

[schnittchen]

Pushed terser version.

[coveralls]

Coverage increased (+0%) when pulling cdd06b0 on schnittchen:dry-run into a04fb17 on rspec:master.

[JonRowe]

Nice work on the compacting, although I'd be tempted to use one liners for the hooks; Also this should now be "1 example"

[schnittchen]

Oh my. Blanked on that. Will fix it tonight.

[coveralls]

Coverage increased (+0%) when pulling 8ac274e on schnittchen:dry-run into 505939b on rspec:master.

[JonRowe]

LGTM, but we'll wait until @myronmarston returns to merge this I think

[schnittchen]

Cool! Let me know if you want me to squash the commits.

[myronmarston]

Thanks for your contribution, @schnittchen!

I merged it, and then built off of this to add a few further improvements: #1056.