- New type pattern
- Box, Arc, Rc, Vec, Option
FromIterator
cannot be implemented forOption<T>
so#[knuffel(child, unwrap(_))]
approach can't be used
- Semantic difference between structs with and without fields (unit/tuple structs)
- Enum variants with single
Decode
element without attribute vs variants with single scalar element with attribute - Do we need
arguments
/properties
/children
? Is there a case where a type can be used for both patterns? - Lack of documentation about tuple structs
- Enum variants with single
#[knuffel(child, unwrap(children(name)))]
plugins: Vec
instead ofVec
doesn't work?
We only use [
]
as quantifier not {
}
like [n]
, [m,n]
.
In expression [m,n]
, m
or n
can be omitted like [,n]
, [m,]
.
And you can't use *
, +
, and ?
as quantifier anymore (first restriction).
*
is used as any single character with quantifier instead of .
. For example *[1]hoge
matches ahoge
, bhoge
, choge
.
Without quantifier, *
operates as a wildcard, which means it is interpreted as *[0,]
rather than *[1]
, so you can express partial matches like *hoge*
, hoge*
(forward match) and *hoge
(backward match). Now you can't express them like ^hoge
and hoge$
anymore (second restriction).
export LC_ALL=C.UTF-8
export LANG=C.UTF-8
I bring a strong proposal that is prematurely rejected but indeed the most likely to be accepted to our community. That is to select as
spelling not :=
, and modify with
and except
semantics at the same time. For that, we need to investigate what will be derived from the change throughly and here are the results.
This alternative has already been out of consideration, PEP 572 says, but is still here because many Python developers are wondering if as
spelling can go and there are few discussions about the way of modifing the two semantics. Even PEP 572 shows little attension to this possibility.