(MODULES-2957) Fix registry_value tests #143
Conversation
|
Running through adhoc |
87b7bf8 to
6476afd
Compare
|
Adhoc passed. Ready for merge |
| @@ -0,0 +1,13 @@ | |||
| class mixed_default_settings { | |||
There was a problem hiding this comment.
I'm not sure what mixed_default_settings means in this context?
There was a problem hiding this comment.
mixes default value_name and non-default value_name (default as in default registry key value
'hklm\Software\foo\\' vs 'hklm\Software\foo'
There was a problem hiding this comment.
I'll add comments into the test to explain this.
There was a problem hiding this comment.
Add an ASCII art reg tree to show what's going on
| expect(subject.autorequire).to eq([]) | ||
| end | ||
|
|
||
| it 'should only autorequire the immediate ancestor registry_key resource' do |
There was a problem hiding this comment.
This test is slightly wrong. Autorequries grabs the first nearest parent key
| it 'should only autorequire the immediate ancestor registry_key resource' do | ||
| catalog.add_resource(Puppet::Type.type(:registry_key).new(:path => 'hklm\Software', :catalog => catalog)) | ||
| catalog.add_resource(Puppet::Type.type(:registry_key).new(:path => 'hklm\Software\foo', :catalog => catalog)) | ||
| catalog.add_resource(Puppet::Type.type(:registry_key).new(:path => 'hklm\Software\foo\bar', :catalog => catalog)) |
There was a problem hiding this comment.
Add one new test here like:
catalog.add_resource(Puppet::Type.type(:registry_key).new(:path => 'hklm\Software\foo\bar\baz\ethan', :catalog => catalog))And make sure that it autorequires nearest value of hklm\Software\foo\bar rather than non-existant hklm\Software\foo\bar\baz which doesn't exist.
79bbda1 to
563c54e
Compare
|
Ready for merge |
Previously the tests for the registry_value type were incomplete or were not testing what they said they were. This commit; * Adds a compile test for when one resource is setting a default value and another is not, but shares the same name. A fix for this was implemented in 64bba67 however there were no tests for this behaviour * Fixed a bug in the canonicalising of path names which would append a backslash every time the method was called * Actually tested that multiple backslashes are removed on the path * Raised errors when an non default, but root only path was passed into a registry value. This is different to registry keys, where this is allowed. * Added tests for autorequiring of registry_key from registry_value resources * Added tests in the registry_value provider for creating and destroying default values * Added registry_value provider tests for case-insensitive exists? method and added for creating and destroying default values * Fixed typo in the registry_key type tests whereby it was actually testing the resource aliases, not the autorequire of the type
563c54e to
9781761
Compare
| # | | ||
| # + foo | ||
| # + (default value) <-- This is the default value for a key called 'HKLM\Software\foo'. | ||
| # This is the 'hklm\Software\foo\\' resource |
| @@ -60,12 +61,55 @@ | |||
| it 'should validate the length of the value data' | |||
| it 'should be case-preserving' | |||
| it 'should be case-insensitive' | |||
There was a problem hiding this comment.
We can worry about this later.. but any reason to leave these since they don't do anything either?
| @@ -51,7 +51,12 @@ | |||
|
|
|||
| describe "#exists?" do | |||
| it "should return true for a well known hive" do | |||
| reg_value = type.new(:path => 'hklm\SOFTWARE\Microsoft\Windows NT\CurrentVersion\SoftwareType', :provider => described_class.name) | |||
| reg_value = type.new(:title => 'hklm\SOFTWARE\Microsoft\Windows NT\CurrentVersion\SoftwareType', :provider => described_class.name) | |||
There was a problem hiding this comment.
All the other tests in this file use :path rather than :title (given :path is the namevar they're currently equivalent) ... I assume this ties into the next PR? Just a little peculiar to see a different attribute for the type without any explanation for why :title is used here rather than :path.
There was a problem hiding this comment.
copy-pasta typo. Should've been path
|
A couple minor comments we can look at later - good enough for now given other changes pending. |
Previously the tests for the registry_value type were incomplete or were not
testing what they said they were. This commit;
another is not, but shares the same name. A fix for this was implemented
in 64bba67 however there were no tests for this behaviour
every time the method was called
registry value. This is different to registry keys, where this is allowed.
values
added for creating and destroying default values
resource aliases, not the autorequire of the type