I don’t think “asymmetric” means anything formal here. A more rigorous explanation would be that the guidance on (quoted below) doesn’t generalize to all arities.

Note how the transient-using version has exactly the same structure, just:

- Calling transient on the source vector

- Using conj! instead of conj

- Calling persistent! on return

That is, if you’re using a higher arity of conj, you need to make multiple calls to conj! instead of simply adding the exclamation point.

The description and patches actually does several things at once:

• Add the above arity generalization for variadic conj! and arity 1 (dissoc! m)

• Add an arity-1 base case to assoc.

  • To me, this is the main feature I’m interested in of this patch. I’ve encountered bugs in the wild due to the expectation that (apply assoc m xs) is valid for an empty xs.

  • Makes the same generalization for assoc

• Adapt the patch for assoc!

It might be easier to evaluate the assoc arity-1 patch independently, which basically boils down to: this:

“The function arities and behaviors between conj, assoc, disj, and dissoc are asymmetric to their transient versions.” What does that sentence mean?

No worries, and I thank you for keeping the patch current, even while keeping my name on it.

That was not intentional at all, thanks for catching it, I'm honestly not sure what happened on my end to make that happen. I think this patch should be a correct update. Sorry about that.

Is it intentional that clj-1103-9.patch leaves out the changes to assoc! that were in the earlier patches? I can create another updated one if so – just curious if there was a reason for the change.