Re: The map-lookup-forgiveness problem

Submitted:

"[XPath] Error-free selection operator for maps or arrays, or finite-domain
functions"

https://github.com/qt4cg/qtspecs/issues/341



On Tue, Feb 7, 2023 at 3:34 PM Dimitre Novatchev <dnovatchev@gmail.com>
wrote:

> >  What's the objection to writing `E[. instance of map(*)]?X ` - is it
> just that's it's wordy?
>
> In case we find a short and intuitive operator-like string, I am all for
> it!
>
> On Tue, Feb 7, 2023 at 1:36 PM Michael Kay <mike@saxonica.com> wrote:
>
>> The null-coalescing operators are subtly different. With the expression
>> E?X we already allow E to be an empty sequence, so in that sense it is
>> already null-coalescing.
>>
>> The question we're addressing, if I understand it correctly, is what to
>> do when E is a sequence, and when it contains some items for which e?X is
>> defined (i.e. maps and arrays) and other items for which e?X is not defined
>> and throws an error.
>>
>> There's a proposal to define ?? as a (kind of) transitive closure of ?
>> (?? is to ? as // is to /, roughly), and I would certainly expect that to
>> only go down the routes that are applicable, ignoring items that are not.
>> If it finds a leaf node containing a string or number, it won't try to
>> apply "?" to it recursively (obviously).
>>
>> What's the objection to writing `E[. instance of map(*)]?X ` - is it just
>> that's it's wordy?
>>
>> Michael Kay
>> Saxonica
>>
>> > On 7 Feb 2023, at 21:15, Liam R. E. Quin <liam@fromoldbooks.org> wrote:
>> >
>> > On Tue, 2023-02-07 at 12:29 -0800, Dimitre Novatchev wrote:
>> >> Addressing the "difficulty to debug" issue, this reminds me of C#
>> >> where
>> >> there are both the "." (member access)  and "?."  (null-conditional)
>> >>   operators.
>> >>
>> >>
>> https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/operators/member-access-operators#null-conditional-operators--and
>> >> -
>> >>
>> >
>> > This has recently been added to JavaScript too as i understand it.
>> >
>> > There is also the null-condensing operator ?? in which a ?? b is like
>> > our (a, b)[1]. So i think we should avoid using ?? to mean something
>> > similar but subtly different; || is bad enough already :)
>> >
>> > JavaScript has ?. as null-safe property access, and this has widespread
>> > support. See
>> >
>> https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Optional_chaining#browser_compatibility
>> > function printCustomerCity(customer) {
>> >  const customerCity = customer?.city ?? "Unknown city";
>> >  console.log(customerCity);
>> > }
>> >
>> > which uses both of these operators.
>> >
>> > As a side-note, there are also nullish assignment operators, e.g.
>> >  a ??= e;
>> > is like XPath's
>> >  a := if (a) then a else e
>> > except we can't update variables of course, so this is pernicious
>> > shadowing...
>> >
>> > and ||= and &&=, neither of which we could use :)
>> >
>> > But i think we probably should avoid using ?? to mean the same as
>> > JavaScript and C# ".?" - this will especially confuse SaxonJS users :)
>> > and lead to bugs that are hard to find.
>> >
>> > I do however think a safe dereference operator is a good idea, and if
>> > it wasn't for JS using ?? in that way, i'd like ?? for it.
>> >
>> > Maybe we should use ‽ instead :) or allow [.]? to work that way, so
>> >  e1[.]?e2
>> > returns () if e1 is empty, e1?e2 if e1 is a map, and an error
>> > otherwise.
>> >
>> > liam
>> >
>> > --
>> > Liam Quin, https://www.delightfulcomputing.com/
>> > Available for XML/Document/Information Architecture/XSLT/
>> > XSL/XQuery/Web/Text Processing/A11Y training, work & consulting.
>> > Barefoot Web-slave, antique illustrations:  http://www.fromoldbooks.org
>> >
>>
>>
>>
>
>