Add --exclude_pattern option to configuration #1651
Conversation
| When I run `rspec` | ||
| Then the output should contain "3 examples, 0 failures" | ||
|
|
||
| Scenario: The `--exclude-pattern` flag makes RSpec first match the default pattern, then skip files matching the specified exclude pattern |
myronmarston
Jul 28, 2014
Member
The wording on this ("...makes RSpec first match the default pattern...") makes it sound like --exclude-pattern can't be combined with --pattern. If that's true, we should find a way to make them work together. If it's not true, we should find a better wording for this :).
The wording on this ("...makes RSpec first match the default pattern...") makes it sound like --exclude-pattern can't be combined with --pattern. If that's true, we should find a way to make them work together. If it's not true, we should find a better wording for this :).
jmondo
Jul 28, 2014
Author
Contributor
Ah you're right. They do work together. I'll work on that wording.
Ah you're right. They do work together. I'll work on that wording.
myronmarston
Jul 28, 2014
Member
Also, maybe we can add a scenario showing them working together?
Also, maybe we can add a scenario showing them working together?
jmondo
Jul 28, 2014
Author
Contributor
Added.
Added.
| @@ -1,7 +1,7 @@ | |||
| Feature: pattern | |||
|
|
|||
| Use the `--pattern` option to tell RSpec to look for specs in files that match | |||
| a pattern other than `"**/*_spec.rb"`. | |||
| a pattern instead of the default `"**/*_spec.rb"`. | |||
myronmarston
Jul 28, 2014
Member
👍 This reads better, thanks.
| @@ -193,6 +193,19 @@ def pattern=(value) | |||
| @pattern = value | |||
| end | |||
|
|
|||
| # @macro define_reader | |||
| # Load files except those matching this pattern. | |||
myronmarston
Jul 28, 2014
Member
This wording suggests RSpec will load every file on the file system that does not match the specified pattern. Maybe this is better?
# Excludes files that match the specified pattern from being loaded by RSpec.
This wording suggests RSpec will load every file on the file system that does not match the specified pattern. Maybe this is better?
# Excludes files that match the specified pattern from being loaded by RSpec.
jmondo
Jul 28, 2014
Author
Contributor
I think that's clearer. Won't match the wording of pattern though. Is that ok? Pattern:
# Load files matching this pattern (default: `'**/*_spec.rb'`))
I think that's clearer. Won't match the wording of pattern though. Is that ok? Pattern:
# Load files matching this pattern (default: `'**/*_spec.rb'`))
myronmarston
Jul 28, 2014
Member
It would be nice to use similar warning but I care more about accuracy and clarity. Any ideas for a consistent wording that is clear and accurate for both?
It would be nice to use similar warning but I care more about accuracy and clarity. Any ideas for a consistent wording that is clear and accurate for both?
jmondo
Jul 28, 2014
Author
Contributor
"Load files matching this pattern" (existing)
"Exclude files matching this pattern" might work.
"Load files matching this pattern" (existing)
"Exclude files matching this pattern" might work.
jmondo
Jul 28, 2014
Author
Contributor
Changed it to this, let me know if you would prefer something else. Seems succinct and not misleading.
Changed it to this, let me know if you would prefer something else. Seems succinct and not misleading.
| When I run `rspec --exclude-pattern "**/{models,features}/*_spec.rb"` | ||
| Then the output should contain "0 examples, 0 failures" | ||
|
|
||
| Scenario: The `--exclude-pattern` flag can be used with the `--pattern` flag |
jmondo
Jul 28, 2014
Author
Contributor
The new test!
The new test!
| assign_files_or_directories_to_run file | ||
| expect(config.files_to_run).to include(file) | ||
| end | ||
|
|
myronmarston
Jul 28, 2014
Member
Why did you remove this spec?
Why did you remove this spec?
jmondo
Jul 28, 2014
Author
Contributor
Ah right meant to comment about that. It wasn't testing anything. Take a look at https://github.com/jmondo/rspec-core/blob/exclude-pattern/lib/rspec/core/configuration.rb#L1289 . When you run rspec with a file specified (not a directory), it loads that file rather than doing the pattern matching. Try changing "/resources/a_foo.rb" in the test on 562 to one of the other specs in the resources folder and it should still pass.
Ah right meant to comment about that. It wasn't testing anything. Take a look at https://github.com/jmondo/rspec-core/blob/exclude-pattern/lib/rspec/core/configuration.rb#L1289 . When you run rspec with a file specified (not a directory), it loads that file rather than doing the pattern matching. Try changing "/resources/a_foo.rb" in the test on 562 to one of the other specs in the resources folder and it should still pass.
cupakromer
Jul 28, 2014
Member
I'll agree this is a slightly poor spec. Though, it is testing something. It's just not explicitly clear from the spec name. This is likely a duplicate of the spec on line 408. The name of spec 408 is clearer, however, it never sets a pattern, so it's arguably incomplete - unless it was intended to by-pass the setting of a default pattern?
Once we sort out if these are dupes, or should be testing slightly different things, I think it's possible to improve the description to clear up any confusion.
I'll agree this is a slightly poor spec. Though, it is testing something. It's just not explicitly clear from the spec name. This is likely a duplicate of the spec on line 408. The name of spec 408 is clearer, however, it never sets a pattern, so it's arguably incomplete - unless it was intended to by-pass the setting of a default pattern?
Once we sort out if these are dupes, or should be testing slightly different things, I think it's possible to improve the description to clear up any confusion.
myronmarston
Jul 28, 2014
Member
It is a weird spec, but I think there's value in having a spec that shows what happens when users pass a specific file name (rather than a directory), and how that relates to pattern. It looks like the behavior is that the pattern is ignored and the file is loaded since it was explicitly specified. I think it would be useful to replace this with a spec that calls assign_files_or_directories_to_run with two file names -- one that matches the pattern and one that doesn't -- and have it show that both files are included in files_to_run, demonstrating that the pattern is not applied to specific file names.
This makes me realize we should probably also spec how exclude_pattern relates to specific files...seems like it is most consistent to do the same thing: specific files are always loaded regardless of whether or not they match either pattern or exclude_pattern.
Does that make sense?
It is a weird spec, but I think there's value in having a spec that shows what happens when users pass a specific file name (rather than a directory), and how that relates to pattern. It looks like the behavior is that the pattern is ignored and the file is loaded since it was explicitly specified. I think it would be useful to replace this with a spec that calls assign_files_or_directories_to_run with two file names -- one that matches the pattern and one that doesn't -- and have it show that both files are included in files_to_run, demonstrating that the pattern is not applied to specific file names.
This makes me realize we should probably also spec how exclude_pattern relates to specific files...seems like it is most consistent to do the same thing: specific files are always loaded regardless of whether or not they match either pattern or exclude_pattern.
Does that make sense?
jmondo
Jul 30, 2014
Author
Contributor
Yep. Working on it now.
Yep. Working on it now.
| end | ||
|
|
||
| context "after files have already been loaded" do | ||
| it 'will warn that it will have no effect' do |
myronmarston
Jul 28, 2014
Member
Not a merge blocker, but I'd prefer it 'warns....' rather than it 'will warn...'. It's more inline with our phrasing and is more terse while being just as clear.
Not a merge blocker, but I'd prefer it 'warns....' rather than it 'will warn...'. It's more inline with our phrasing and is more terse while being just as clear.
| config.exclude_pattern = "rspec/**/*.spec" | ||
| end | ||
|
|
||
| it 'will not warn if reset is called after load_spec_files' do |
myronmarston
Jul 28, 2014
Member
Same here.
Same here.
|
BTW, if there reaches a point where you don't have time or desire to continue working on this, let us know and we'll be happy to pick it up. We like to let contributors finish there PRs when they are able, though :). |
|
Good to know. I'm planning on finishing it! Let me know if I need to hurry up. |
|
Actually feel free to make the changes required to support those other two merges. Let me know if there's anything I should do with my feature after that. |
|
OK. I've got a couple other PRs on my list to tackle first, but I'll add yours to my list. In the meantime, if you want to take this further, feel free :). |
|
I'm going to pick this up in the next couple days and get it ready to merge. |
See #1651 (comment) for background discussion.
|
Closing in favor of #1671. |
See rspec#1651 (comment) for background discussion.
Resolves #1626. It allows passing
--exclude_patterntorspecon the command line.Afaik we still can't use the power of rake tasks. This is because the rake task still uses
FileListto include files. And once files are included, they can't be changed by passing--patternor--exclude-pattern(there is a warning about this when you try). Fortunately, once #1589 is merged, it will be possible to pass--exclude-patternwithinrspec_optsin the rake task config. This is because the rake task'spatternoption will use--patternwhich can be combined with--exclude-patternfor a one-two punch.@myronmarston Let me know how this looks! Sorry it took so long.