Version 9 improvements

[broofa]

This started out as a variation of #463 (uuidToBytes implementation), but now includes the following per discussion in this PR:

• Rename uuidToBytes() -> stringify()
• Rename bytesToUuid() -> parse()
• parse() (was uuidToBytes()) now returns Uint8Array
• Expose version(), validate(), parse(), and stringify() as part of public API
• Unit tests and benchmarks for stringify() and parse()
• Document new methods in README

[broofa]

Pushing forward with this approach. I can move relevant work to #463 if we decide to go with that.

• Pulled uuidToBytes into separate file
• uuidToBytes unit test
• Added uuidToBytes to exports
• Added benchmark for uuidToBytes.

[ctavan]

LvGTM (once the minor nits in the test + naming are solved).

[broofa]

Comments look good, @ctavan. I'm out of town until wednesday, however, so may not be able to get to this until then.

[broofa]

Adding @LinusU and @TrySound as reviewers, as there's been a bit of feature creep here.

Requesting re-review from @ctavan because switching to Uint8Array for parse() has resulted in a few changes elsewhere in the code. Also, it's worth noting that this change reduced the benchmark performance of parse() by ~25%.

[broofa]

@ctavan: Note that the UMD version of the API has uuid prefixed on the new methods. So, uuidVersion, uuidParse, uuidParse(), and uuidStringify(). This is mostly consistent with the existing uuidv1 methods. (The unprefixed names felt a bit too generic to export as globals so... 😕 ). Let me know if you have a better idea for how to handle this.

Edit: Also, what's up with the failing webdriver tests? Is that just browserwatch being flakey?

[ctavan]

Looks good!

A second review by @LinusU or @TrySound would be really welcome!

[ctavan]

@ctavan: Note that the UMD version of the API has uuid prefixed on the new methods. So, uuidVersion, uuidParse, uuidParse(), and uuidStringify(). This is mostly consistent with the existing uuidv1 methods. (The unprefixed names felt a bit too generic to export as globals so... 😕 ). Let me know if you have a better idea for how to handle this.

Not really a better idea. Consistent would be uuidversion but that doesn't read well… I consider the umd build really just an optional feature e.g. for educational purposes where people do online coding courses or things like that, so I'm not so much afraid about the api surface on that end.

Also, with cool things like jspm there seems to be a modern alternative for the direct browser loading use case.

So maybe just not expose these methods in the umd build at all until somebody complains?

Edit: Also, what's up with the failing webdriver tests? Is that just browserwatch being flakey?

Test fail in the old browsers:

• IE11
• Firefox 44
• Chrome 49

String.padStart() seems to be the culprit:

1592943821280:SEVERE:http://127.0.0.1:9000/browser-webpack/dist/all.js 1:991 Uncaught TypeError: o.toString(...).padStart is not a function
,1592943821339:SEVERE:http://127.0.0.1:9000/favicon.ico 0:0 Failed to load resource: the server responded with a status of 404 (Not Found)
,1592943831873:SEVERE:http://127.0.0.1:9000/browser-webpack/dist/v1.js 1:992 Uncaught TypeError: o.toString(...).padStart is not a function
,1592943842443:SEVERE:http://127.0.0.1:9000/browser-webpack/dist/v4.js 1:992 Uncaught TypeError: o.toString(...).padStart is not a function
,1592943852840:SEVERE:http://127.0.0.1:9000/browser-rollup/dist/all.js 1:465 Uncaught TypeError: u.toString(...).padStart is not a function
,1592943863391:SEVERE:http://127.0.0.1:9000/browser-rollup/dist/v1.js 1:465 Uncaught TypeError: u.toString(...).padStart is not a function
,1592943873923:SEVERE:http://127.0.0.1:9000/browser-rollup/dist/v4.js 1:461 Uncaught TypeError: r.toString(...).padStart is not a function

[broofa]

@ctavan Where are you seeing those webdriver error messages? I'm not seeing anything like that in the github "checks" logs (e.g. https://github.com/uuidjs/uuid/pull/464/checks?check_run_id=801127791). All I see are timeout errors.

[ctavan]

I got those console errors from the Browserstack UI. Last time I think I had issues inviting you to my Open Source team there. Shall I give that another try?

Seems like now it's Chrome 49 and IE 11 that are broken, however this time I don't get meaningful console errors from Browserstack.

However this screenshot:

tells me that something in v5() must be broken because all lines up until

uuid/examples/browser-webpack/example-all.js

Line 25 in b51f172

addTest('uuidv5() DNS', uuidv5('hello.example.com', uuidv5.DNS));

seem to render fine.

[broofa]

Temporary browserstack workaround: Building and pushing to my personal site, then doing manual testing there. It's not great but I've figured out the immediate issue (No Array.from on IE11). All checks passing.

[LinusU]

LGTM 👍