Feature or enhancement

This is a follow-up to #134696. After c6e63d9, it's now possible to use the following forms:

• hashlib.new(name, data)
• hashlib.new(name, data=...)
• hashlib.new(name, string=...)

and, taking MD5 as an example, hashlib.md5(data), hashlib.md5(data=...) and hashlib.md5(string=...).

In the docs we only expose the following signatures:

hashlib.new(name, [data, ]*, usedforsecurity=True)
hashlib.md5([data, ]*, usedforsecurity=True)

So, it would make sense to remove support for string=.... Ideally, I want to make it positional-only, but I don't think it's worth the shot and it's too much for a breaking change.

Note that we could also deprecate data itself because originally, PEP-247 and PEP-452 were using string and not data, but this is probably confusing as the PEP says:

Although the parameter is called ‘string’, hashing objects operate on 8-bit data only. Both ‘key’ and ‘string’ must be a bytes-like object (bytes, bytearray…).

I assume that the docs decided to use 'data' to remove this ambiguity as 'data' is more common for bytes-like objects.

cc @gpshead

Linked PRs

• gh-134978: deprecate string keyword parameter for hash function constructors #134979