> From: Lassi Kortela <xxxxxx@lassi.io>
> Date: Sunday, August 16, 2020 1:31 AM
>
>> I also don't like set, but consider it a placeholder for now, we should
>> be able to come up with a better name.
>>
>> But before going to that effort, we have to agree on the need for it.
>> If it's not mandatory, it's a */very/* different thing, and while
>> "convention" is a synonym for me in context, "protocol" isn't particularly.
>
> IMHO the purpose of 'set is to give meaning to 'code/number and
> 'name/symbol. 'code/number and 'name/symbol make sense without 'set, in
> the same way that a grammatically correct but taken-out-of-context
> sentence makes sense in English. It refers to something definite but
> since you lack the context, you don't know what specifically.
>
> In that sense 'set is not technically mandatory yet is obviously useful.

Thus we have incompatible visions for SRFI 198.

> What are we really trying to get to the bottom of -- is this about the
> error codes that are non-specific negative integers, where we mainly
> know that negative means error and the precise meaning of a particular
> number may be unknown?

In my view, that sort of thing should be encapsulated by the implementor
using SRFI 198.  E.g. for libsodium, I use the SRFI 170 paradigm of
silence on success, raise an error on failure.  So the end user never
needs to know it's a classic UNIX(TM) return 0 on success, -1 on failure.

Users who return statuses rather than raise error would presumably use
the standard defined key we've previously discussed, who's name
escapes me at the moment ('success? ??).

>>> We're gonna need a reflection API at some point (i.e. enumerating sets
>>
>> What are going to be the consumers of this reflection API?
>
> Mostly Scheme programmers using a REPL or IDE.

OK; what's their use cases for doing reflection, to in your example
find all the available 'sets in their current environment?  Why would
they need that information?

- Harold