Automatically detect implementation languages on GitHub #1203
Conversation
If a project is created, and its repo is on GitHub, use the GitHub API to discover the implementation languages in use and fill in that information. This required updating several vcr_cassettes. Signed-off-by: David A. Wheeler <dwheeler@dwheeler.com>
Codecov Report
@@ Coverage Diff @@
## master #1203 +/- ##
======================================
Coverage 100% 100%
======================================
Files 47 47
Lines 1684 1693 +9
======================================
+ Hits 1684 1693 +9
Continue to review full report at Codecov.
|
| :"C#" => 29_834, | ||
| HTML: 2_290, | ||
| Ruby: 1_359 | ||
| ) |
There was a problem hiding this comment.
This seems like a bad thing to build into a test, since the values will change the next time the cassette is re-recorded, right?
Could you instead run on a known value that won't change, like the values of a specific release: https://github.com/kubernetes/kubernetes/tree/release-1.0
There was a problem hiding this comment.
Oh wait - these values will not change when the cassette is re-recorded. The hash value here is sent to a specific method which reformats them from a string - it doesn't retrieve anything from GitHub. The JSON values here were originally from assimilation, but I just used those because it's a real-world complicated list. So it turns out that it's not a problem here!
|
LGTM if you address my one comment. |
|
I'm not sure fixing the version is better. I'd like the test to be like the real thing as much as possible. I think it is highly unlikely for these values to change anyway; if they do, we can change the values to match. Also, the mismatch can only occur when we recompute cassettes, which is rare. |
|
Fixing the version to analyze does not relly fix the results anyway; this is a service request, and GitHub sometimes updates the heuristics used in their analysis to determine languages. |
|
Simplest would be for ciitest to fork a repo (like the best practices one) and then not change it. That should be relatively stable. |
|
Yes, I think that's the better approach. Then the test will be more realistic. Changes could still happen, but they would be at least relatively rare. |
|
OK, please merge then.
…--
Dan Kohn <dan@linuxfoundation.org>
Executive Director, Cloud Native Computing Foundation https://www.cncf.io
+1-415-233-1000 https://www.dankohn.com
On Sun, Aug 5, 2018 at 11:21 PM, ciitest ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In test/unit/lib/github_basic_detective_test.rb
<#1203 (comment)>
:
> + assert_equal 'C#, JavaScript', detective.language_cleanup(
+ HTML: 1000, :"C#" => 5000, JavaScript: 800
+ )
+ # This is the list from https://api.github.com/repos/
+ # assimilation/assimilation-official/languages
+ assert_equal 'Python, C, Shell, C++, CMake, C#, Ruby',
+ detective.language_cleanup(
+ Python: 1_151_127,
+ C: 1_059_779,
+ Shell: 285_358,
+ :"C++" => 44_086,
+ CMake: 33_855,
+ :"C#" => 29_834,
+ HTML: 2_290,
+ Ruby: 1_359
+ )
Oh wait - these values will *not* change when the cassette is
re-recorded. The hash value here is sent to a specific method which
reformats them from a string - it doesn't retrieve anything from GitHub.
The JSON values here were originally from assimilation, but I just used
those because it's a real-world complicated list. So it turns out that it's
not a problem here!
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#1203 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AC8MBlV7s6fC-0XWNsWdfENoQKFoAsc2ks5uN7YpgaJpZM4Vu6Y0>
.
|
|
That test already uses some data from a ciitest project, but it didn't test the detected languages - because there weren't any!! I have added a Python file to the ciitest project, so we can have a more realistic test, and keeps the unit tests that try out other options. I'll push that version up soon. So that resolves everything. Thanks! |
This adds a direct test of the detected implementation languages for the simple ciitest test project. Signed-off-by: David A. Wheeler <dwheeler@dwheeler.com>
If a project is created, and its repo is on GitHub, use the
GitHub API to discover the implementation languages in use
and fill in that information. This required updating
several vcr_cassettes.
Signed-off-by: David A. Wheeler dwheeler@dwheeler.com