- From: Miller, Bruce R. (Fed) <bruce.miller@nist.gov>
- Date: Sat, 9 Oct 2021 07:40:54 -0400
- To: Neil Soiffer <soiffer@alum.mit.edu>
- Cc: "www-math@w3.org" <www-math@w3.org>

On 10/9/21 12:10 AM, Neil Soiffer wrote: > I think I must not be getting your gist in: > > But if there are contexts & defaults applied, you have to make sure > that x' is *not* treated as a derivative (by whatever means). > > The point is that we not only have to provide a way to assert what > something *is* (or how it should be pronounced), but in the presence > of defaulting, we essentially have to say what it's *not*. > > > The point of having defaults is precisely so that you know when adding intent is required > to convey some specific meaning that differs from the default. If software doesn't know > what a default is, then it should generate an intent value. That's exactly what makes using defaults a bit tricky: To annotate a document accurately, you'll need to know exactly which rules will be applied to know whether or not they will need to be overruled. > In case it isn't obvious, defaults should be simple, but not necessarily trivial. For > example, I think saying "msup" defaults to power is too trivial, but a simple rule like > "msup is a power except when the base is an <mi> and exponent is an <mo> , in which case > it is an identifier name." (feel free to pick holes in this -- it is just meant to be an > example showing that one can add a few special cases to a default to make it more useful. > This doesn't help for derivative, or transpose, or many other notations, but it does cover > x^*, the x^' examples in the math counts problems, along with it being a power. Right. But I'm thinking that it would be perfectly reasonable for, say, a calculus or dynamics context's rule set to say that <msup> <anything> <mo>prime</mo> </msup> gets a default intent of derivative (in whatever syntax). That would match *both* the f' and x' in the example. > Neil > > > > On Fri, Oct 8, 2021 at 1:56 PM Miller, Bruce R. (Fed) <bruce.miller@nist.gov > <mailto:bruce.miller@nist.gov>> wrote: > > On 10/8/21 3:00 PM, Neil Soiffer wrote: > > To return to some of the examples. If someone uses a prime, double prime, etc, a > literal > > reading "x double prime" doesn't need anything special > > Deyan addressed the question of what a "natural" reading of primes may or may not be. > > I want to emphasize a different point about specialized pronunciation & "defaults". > > In real life, I've encountered things like "f' = x' + ....", > where for example f is a function and prime means to take its derivative, > while x' is a unique variable, presumably some transformation of x. > > If we *never* are going to concern ourselves with output other than the literal: > "eff prime equals ecks prime dot dot dot" > and leave it to the hearer to figure it out, then there's no issue. > (but that seems not to be your POV) > > If, on the other hand, we are expecting to say something different for > a derivative (whatever it might be), then the two instances will *have* > to be distinguished. > If there were no defaulting, we would simply assert that f' is a derivative. > (but that seems not to be the POV of several folks) > > But if there are contexts & defaults applied, you have to make sure > that x' is *not* treated as a derivative (by whatever means). > > The point is that we not only have to provide a way to assert what > something *is* (or how it should be pronounced), but in the presence > of defaulting, we essentially have to say what it's *not*. > > bruce > -- > bruce.miller@nist.gov > http://math.nist.gov/~BMiller/ > -- bruce.miller@nist.gov http://math.nist.gov/~BMiller/

Received on Saturday, 9 October 2021 11:41:12 UTC