Request to make validation optional for Stringify OR export unsafeStringify

[vineet-sh]

Previously I was using bytesToUuid (v3.4.0) which I understand is now stringify, but uuid validation in stringify is invalidating most of uuids. It would be great if we can do any of the following:

• Either export the unsafeStringify as a separate API
• Or make validation part in stringify as optional

[LinusU]

[...] is invalidating most of uuids.

What do you mean by this? 🤔

Are you trying to stringify a byte array that is not a uuid?

[vineet-sh]

Yes, those were unconstrained 128 bit numbers which were converted to uuid like format using bytesToUuid.
Presently, stringify validates those against uuid regex and invalidates them, which is although the correct behavior, but it would be great if bytesToUuid (presently unsafeStringify) functionality can be restored along with the present APIs

[broofa]

Either export the unsafeStringify as a separate API

My rationale for why we shouldn't do this. tl;dr: it increases the maintenance burden.

Or make validation part in stringify as optional

I appreciate that bytesToUuid() used to take any 16-byte array, but uuid is no longer in the business of generating or otherwise working with invalid UUIDs. That it complains loudly and early upon encountering a bad UUID is, to my mind, a good thing. It allows us to keep the code and public API focused, and helps users quickly identify issues with upstream code. (Instead of, for example, allowing stringify() to generate invalid UUIDs that might propagate into DB tables, URLs, etc.)

[vineet-sh]

Got it, thanks :)