Aram Panasenco

Thank you @triska! I think this representation is sufficient for practical JSON use - JSON objects with more than about 20 keys are highly abnormal in my experience. Declaratively, the...

That's interesting, I didn't know a partial list was excluded from the definition of a list! In that case, would it be a best practice to use `catch(must_be(list, OrdPairs), error(instantiation_error,_),...

I'm on board with instantiation errors, however the problem is that raising an instantiation error in this DCG would render it unusable for parsing. I'll have to give this some...

Using `string_si/1` from #943, I believe the below definition for `uniquestringkeysortedlist_si/1` would work to check whether a variable is a list of pairs sorted by unique string keys soundly (ignoring...

OK @triska, take a look at `json_object//1` now and tell me whether I succeeded in preserving inference soundness while also retaining the ability to both parse and generate with one...

This is not really relevant to the PR, but I'm just very excited that schemas are slowly starting to become possible with JSON objects as keysorted lists: ``` ?- Msg...

Whenever you say anything, @triska, my first reaction is "that's impossible!" and then sometime later "that was obvious, why didn't I think of that?" 😅 Take a look at `json_object//1`...

Now we just need `string_si/1` (#943) to be finished first.

While I'm waiting for `string_si/1` to be merged, I also rewrote `json_character//1` in a similar fashion to use `freeze/2` instead of instantiation checks.

Is it possible to put a workaround in the installer that temporarily appends the install prefix to PYTHONPATH so that installation doesn't crash and people don't have to google the...