Extend GitHub source analysis

[ctavan]

I have finally managed to clean up and improve the queries and add some tooling to help everyone reproduce my results.

In contrast to my initial numbers from https://github.com/bcoe/proposal-standard-library-uuid/issues/4 I have changed my analysis in a few ways:

• I am now less strict when collecting the repos to be considered (not only those with top-level package.json files).
• I no longer include repos that use the legacy node-uuid module (which happen to be quite a few @broofa: Almost an equal amount of repos I found use node-uuid compared to uuid…).
• I count each repo where a certain UUID version is being used only once towards the results (before I was stupidly counting lines where uuid.v1/3/4/5 was appearing which skewed results).
• I calculate a weighted popularity score based on repo watch count.

In addition to the global statistics I have assembled a list of the top 100 GitHub repos that make use of each UUID version in this Google Sheet: https://docs.google.com/spreadsheets/d/1NjrsNgEZaXs10tXBRGgMpA-9rh_a3rEQlKfi1TpAnYI

We may use this sheet in order to extract use cases as suggested in #14 and to potentially get in touch with authors to discuss why they are using v1 uuids as suggested in https://github.com/bcoe/proposal-standard-library-uuid/issues/4#issuecomment-500823549.

Let me know what you think!

[broofa]

What's the "Seems legit?" column in the table? (Also, consider renaming that column since putting "No" there implies a project is "not legit", which is probably not a message we want to be sending.)

[ctavan]

Oh, with "Seems legit?" I meant whether UUID v1 usage seems a reasonable choice in those projects from my point of view (I didn't mean to question the legitimacy of the repos themselves).

Renamed to "v1 usage reasonable?" and put a "questionable" wherever I believe v4 would be the better choice and where I intend to get in touch with the maintainers.

Feel free to further improve the wording. Since I'm not a native speaker I sometimes have a hard time finding the right "mood"…

[ctavan]

@bcoe @codehag I have finally managed to include my hypotheses into the readme of the analysis and I have also added a reference from the main readme to the analysis.

I think this (especially the READMEs) should be good for a review now.

[ctavan]

@bcoe thanks for your review!

Apart from the question regarding standard deviation I have addressed your comments and also rebased onto master.

Feel free to comment again (or merge).

[ctavan]

@bcoe as this PR has grown quite big already are you fine with merging it for now and then iterate in future PRs as necessary?