On Sat, Feb 15, 2025 at 3:13 PM Peter Saint-Andre <stpeter@xxxxxxxxxx> wrote:
On 2/15/25 3:38 PM, Rob Sayre wrote:
> On Sat, Feb 15, 2025 at 11:34 AM Peter Saint-Andre <stpeter@xxxxxxxxxx
> <mailto:stpeter@xxxxxxxxxx>> wrote:
>
> The PRECIS specifications, which
> followed the lead of the IDNA specifications in this regard, are
> primarily intended to apply to strings that go in protocol slots,
>
>
> Where is "protocol slot" defined? I looked in the PRECIS documents, and
> they do use the term, but I couldn't find a definition.
We never formally defined the term. It is something like "a place
allotted by a protocol specification for communication or exchange of
addresses, identifiers, and other constructs critical to interoperation
between entities in a protocol."
Makes sense. But that implies a container format that PRECIS did not mention either.
Giving the unichars draft a hard time on this point seems a bit harsh. I am surprised that this draft even had to mention PRECIS, since it is at a more basic level (like, it might cover the angle brackets in JSON). This one also covers freeform text, like the body of a tweet/post/etc. So, there can be all sorts of things in there, including private-use characters if we are to accommodate all IRIs (they are allowed in the query string). Or, just total junk as long as they aren't in the "problematic" characters.
I remember we used to allow "unicode vandalist" accounts at Twitter. I tried to find one of those, but only found my own question asking to find it. But it's a decent example:
thanks,
Rob
-- last-call mailing list -- last-call@xxxxxxxx To unsubscribe send an email to last-call-leave@xxxxxxxx
