issue with uuidv5 and/or parse function - Invalid UUID

Describe the bug

A GUID generated in dotnet is failing to parse. We get an error: Invalid UUID. We encountered the issue with v5, but it’s the same error if you pass to the new parse function. I ran git bisect using this test and it fails starting on commit: 0e6c10ba1bf9517796ff23c052fc0468eedfd5f4.

    describe('v5', () => {
       test('should parse', () => {
       const id1 = v5('test', 'de42f6ff-210c-da9c-a30e-1a4ac2743656');
  
       });
    });

How to reproduce

Run this test:

    describe('v5', () => {
       test('should parse', () => {
       const id1 = v5('test', 'de42f6ff-210c-da9c-a30e-1a4ac2743656');
  
       });
    });

Expected behavior

Should produce a v5 uuid.

Runtime

• OS: [macOS, Linux]
• Runtime: [e.g. Node.js, AWS Lambda]
• Runtime Version: [e.g. 12x]

Additional information

[Any other information or comments that you think will help]

[afterthought]

Ok. So I was wrong about it coming from dotnet. It actually originates from an old titanium based mobile app with a handwritten random number generator that just orders the chars/digits into a guid format. Pretty bad. (I did do some testing with the MS Guid.newGuid() method and it seems compliant).

I definitely see that the ID is non-compliant and now we know why. I’m now in the pinch of figuring out how to move forward. Our design/architecture relies heavily on uuidv5 and now I’ve found that I have a mountain of legacy data that won’t be compatible. For now I can pin the library at 8.2. Not a great long-term solution though. On my end I’m going to see if I can convince the product team to fix the uuid generation and ideally have a conversion strategy for older data.

[broofa]

FWIW, I think there’s some misconceptions around the synonymity of Microsoft GUIDs and RFC4122. Lots of articles out there conflating these two terms, often claiming they’re identical. However, Microsoft is pretty clear that GUIDs are allowed to contain any hex digit.

As @ctavan notes, RFC UUIDs have specific requirements for the version and variant fields, whereas GUIDS can have any hex digts there. I.e. The current behavior - where this library considers Microsoft GUIDs that don’t comply with the RFC spec to be invalid - is the intended behavior.