The big win is that it's now possible to test the future parser with the
latest Puppet release. In the past it wasn't possible to test future parser
with anything higher than 3.5.x.
* Removed some tests that no longer apply and/or are redundant
* Only set OS facts when testing OS-specific behaviors. This simplifies the
facts that must be set in the specs and saves running the same tests
several times when the results wouldn't differ by OS anyway.
Location resources with internal == true are valid, but the module does
not allow them unless one or more of www_root, proxy, location_alias,
stub_status, fastcgi, or location_custom_cfg are set
This patch allows locations with none of those but internal == true to
be created.
This is a re-implementation of jfryman/nginx PR-388.
This conversion is done by Transpec 2.3.4 with the following command:
transpec -c "rake spec" spec/classes/ spec/defines/ spec/spec_helper.rb
* 210 conversions
from: it { should ... }
to: it { is_expected.to ... }
* 19 conversions
from: it { should_not ... }
to: it { is_expected.not_to ... }
* 15 conversions
from: == expected
to: eq(expected)
* 15 conversions
from: obj.should
to: expect(obj).to
For more details: https://github.com/yujinakayama/transpec#supported-conversions
- added raw_prepend / raw_append to vhosts & locations, which
adds raw lines to the vhost/location /without/ semicolons
- added location_raw_prepend / location_raw_append to vhost
for passing through to default location
- added spec tests for new parameters
- cleaned up location specs to match new header/body/footer setup
reducing duplicate checks by testing header/footer separately.
- cleaned up whitespace (2 space) in touched files
- used `<%-` in erb templates to allow the use of leading
whitespace, making the logic flow more readable
- Also adds spec tests for recent vhost updates:
- client_body_timeout
- client_header_timeout
- gzip_types
- testing that the first server_name is used when www_to_non_www is true
For some reason these have been broken forever. Travis was failing,
correctly. I have updated the tests to be correct to how the code
actually works.
I would also suggest that if this is merged, we stop merging PRs that
have failed test runs. Tests matter in a module like this.
