-
Notifications
You must be signed in to change notification settings - Fork 3
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
3 changed files
with
66 additions
and
13 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,56 @@ | ||
| [Done] JW: “equavalent” -> “equivalent” | ||
| [Done] ??: “logcial” -> “logical” | ||
| [Done] TK: on p46: I wonder if we should swap the conditions | ||
| JW: why does it make a difference? | ||
| TK: we don’t need to instantiate the const-reference-t again and it can short-circuit | ||
| JW: any objection to switching the order | ||
| [Done] TS: iter_reference_t<iterator> instead of concat-reference-t<...>? | ||
| JW: yes, that should be the same condition | ||
| TS: that is just stylistic | ||
| JW: it uses less exposition-only stuff | ||
| (swappable_with<iter_reference_t<iterator>, iter_reference_t<iterator>> && indirectly_swappable<iterator_t<maybe-const<Const, Views>>> && …) | ||
| [TODO] TK: there is a rewording in the chat avoiding N: | ||
| [TODO] “The exception specfication is equivalent to the logical AND of noexcept(ranges::swap(...)) and the following expression for every integer …” | ||
| [TODO] JW: I think I prefer that because we start with what we are saying | ||
| [TODO] TS: we need to reword that because std::get is actually throwing | ||
| [TODO] TS: we don’t care about std::get throwing here, though | ||
| [TODO] JW: I don’t have a quick solution to that! That will need some work off-line | ||
| [TODO] TS: we could use get_if but that is ugly: using a * and taking the address with addressof. | ||
| [TODO] TS: we could use declval and write it as a full expression | ||
| [TODO] TK: another suggestion in the chat | ||
| [TODO] JW: that would be a pack of types, i.e., we’d need to declval it - OK, now corrected: | ||
| [TODO] “noexcept(ranges::iter_swap(IT, IT)) for IT being lvalue of type iterator_t<maybe-const<Const, Views>> const, for each type in the pack.” | ||
| [TODO] BST: should we use “conjunction” instead of “logical AND” | ||
| [TODO] DK: I prefer logical and because I know what it means; conjunction I keep needing to look up | ||
| [TODO] TK: wording for p4 in the chat: | ||
| [TODO] “Exception specification is equivalent to: | ||
| [TODO] noexcept(ranges::swap(*x, *y)) && ... && noexcept(ranges::iter_swap(its, its)) | ||
| [TODO] where its is a pack of lvalues of type iterator_t<maybe-const<Const, Views>> const respectively.” | ||
| [TODO] HX: in the previous wording all ranges needed to be sized, now the last range doesn’t need to be sized | ||
| [TODO] JW: the last range doesn’t have to be sized because we don’t skip to the end of it | ||
| [TODO] JW: is there an efficient way to get the last element of a pack | ||
| [TODO] SLY: there is a simple implementation in the paper, see https://godbolt.org/z/TG8MhbWqq | ||
| [TODO] TK: on concat-is-bidirectional: we check it twice | ||
| [TODO] TK: we have similar things in the cartesian_view: we could extract the logic for cartesian-is-common | ||
| [TODO] JW: that makes sense | ||
| [TODO] TS: (2.2.4) p5, p6 “offset + steps < …” doesn’t need a static_cast: you can do signed integer comparison with different sizes, i.e., this one doesn’t needs a cast | ||
| [TODO] TS: I think the advance-fwd thing also doesn’t need a static_cast: either it produces the right type or we are in integer land | ||
| [TODO] HX: I think the issue was that the conversion was explicit | ||
| [TODO] TS: only when you are possibly truncating but widening does not | ||
| [TODO] JW: but without the cast we may end up with the wrong type | ||
| [TODO] TS: they are on the same type | ||
| [TODO] TS: the two casts which are required is on the addition to the iterator | ||
| [TODO] JW: so remove for all that are not += and the equivalence a bit further on (-=) | ||
| [TODO] TK: begin()/end() use {} to initialize it: they should use () | ||
| [TODO] JW: we only use {} if we really want to prevent narrowing conversions | ||
| [TODO] TS: on p19: the static_cast isn’t necessary: the conversion is implicit | ||
| [TODO] JW: it also lacks the > | ||
| [TODO] TS: on p34.1: I don’t think it needs a static_cast but it needs to say what type s is. | ||
| [TODO] TS: there is one s total not one s per integer | ||
| [TODO] TS: it is probably easier to just static_cast s | ||
| [TODO] TK: we may overflow: the difference type may not be able to hold that | ||
| [TODO] TS: if the difference type can’t represent the distance between two iterators that is already broken even without using - | ||
| [TODO] TS: on p34.3: the static_cast is unnecessary | ||
| [TODO] JW: same on p36 | ||
| [TODO] TS: that has the same issue with having just one s and it doesn’t actually say what the type is | ||
| [TODO] TS: I don’t see why we say “denotes the difference” instead of using the expression |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters