Date: Wed, 2 Jan 2019 19:19:05 +0000
From: Harald van Dijk <a...@gigawatt.nl>
Message-ID: <6bbdaef7-ac0a-e528-a177-7c51c4ba6...@gigawatt.nl>
| That's very much relevant, that's the whole cause of this discussion!
What I meant was irrelevant was the current specification's definition
of what characters are legal in a function name, rather than the way
that it is specified.
| If function definitions used a WORD token for the function name rather
| than a NAME token,
I agree with Chet - the current spec uses the word "word" rather loosely,
all over the place - sometimes it means token, other times field, other times
...
You'll tie yourself in knots if you try and apply some specific definition to
that term which is to be the same everywhere it appears. It just isnb't.
| What you're describing here is how it's implemented in practice, not how
| it's specified in the standard.
What's actually in the published standard is nonsense.
We have some proposed revised text, but it isn't correct yet. By all means
suggest revised wording so the standard specifies what it should. However
it serves no useful purpose to argue that we must do X (for some obviously
incorrect X) because the proposed new text says so.
kre
