P3016R6 Resolve inconsistencies in begin/end for valarray and braced initializer lists #8544

AlisdairM · 2025-11-17T07:30:59Z

…initializer lists

It is tricky to properly line-break the list of additional headers that contain the range functions, and that list is growing quite large to be displayed as a list in a single sentence. Prefer to render the list as a table instead. Retained the cross-references to the header synopses from the original text, but they look odd in the table format.

eisenwave

Some discrepancies.

Especially make sure that the noexcept(noexcept(...)) contents are correct.

eisenwave · 2025-11-17T14:42:08Z

source/support.tex

-#define @\defnlibxname{cpp_lib_apply}@                             202506L // freestanding, also in \libheader{tuple}, \libheader{type_traits}
+#define @\defnlibxname{cpp_lib_apply}@                             202506L
+  // freestanding, also in \libheader{tuple}, \libheader{type_traits}


I don't see this in the paper, so it should presumably done in a different PR because it's totally unrelated.

This entry is merely reformatted. I think such change isn't necessary for CI (otherwise, CI should have been failing before) and should be dropped.

The reformatting avoids splitting <type_- and traits> across two lines, which is an improvement. But it should be a separate commit.

eisenwave · 2025-11-17T14:44:55Z

source/numerics.tex

 \end{itemdescr}

-\rSec2[valarray.range]{\tcode{valarray} range access}
+\rSec2[valarray.range]{range access}


Suggested change

\rSec2[valarray.range]{range access}

\rSec2[valarray.range]{Range access}

eisenwave · 2025-11-17T14:48:00Z

source/numerics.tex

-\indexlibrarymember{begin}{valarray}%
+\indexlibrarymember{valarray}{begin}%


The old way was correct, the way it's described in our specification style guidelines.

eisenwave · 2025-11-17T14:48:32Z

source/numerics.tex

-\indexlibrarymember{end}{valarray}%
+\indexlibrarymember{valarray}{end}%


Same here, the old style was correct.

eisenwave · 2025-11-17T14:50:53Z

source/support.tex

+    constexpr const E* data() const noexcept;
    constexpr size_t size() const noexcept;     // number of elements
+    constexpr bool empty() const noexcept;
    constexpr const E* begin() const noexcept;  // first element
    constexpr const E* end() const noexcept;    // one past the last element


While the paper doesn't propose any comments in this area, it's inconsistent to have "tutorial comments" on only some of these functions within this block. Maybe a follow-up commit (possibly on a separate PR) could just remove all of those.

I don't think we need code comments like // number of elements telling us that size() returns the number of elements.

I agree with removing them, but it should definitely be a separate PR not as part of this change.

eisenwave · 2025-11-17T14:54:52Z

source/iterators.tex

+  template<class C> constexpr auto
+    begin(C& c) noexcept(noexcept(c.begin())) -> decltype(c.begin());               // freestanding
+  template<class C> constexpr auto
+    begin(const C& c) noexcept(noexcept(c.begin())) -> decltype(c.begin());         // freestanding
+  template<class C> constexpr auto
+    end(C& c) noexcept(noexcept(c.end())) -> decltype(c.end());                     // freestanding
+  template<class C> constexpr auto
+    end(const C& c) noexcept(noexcept(c.end())) -> decltype(c.end());               // freestanding


To be honest, I liked the style in the paper better,
which is

template<class C> constexpr auto begin(C& c) noexcept(noexcept(c.begin())) -> decltype(c.begin());

We usually break after the template-head, and then maybe also break later in the declaration. But I think the <iterator> synopsis is unusual (unique even?) in doing:

template<class C> constexpr auto

i.e. I think it would be move conventional to do:

template<class C> constexpr auto begin(C& c) noexcept(...) -> decltype(...); // freestanding

and if that doesn't fit, then something like:

template<class C> constexpr auto begin(C& c) noexcept(...) -> decltype(...); // freestanding

eisenwave · 2025-11-17T14:58:02Z

source/iterators.tex

+template<class C> constexpr auto crbegin(const C& c) noexcept(noexcept(c.crbegin()))
+  -> decltype(std::rbegin(c));


Suggested change

template<class C> constexpr auto crbegin(const C& c) noexcept(noexcept(c.crbegin()))

-> decltype(std::rbegin(c));

template<class C> constexpr auto crbegin(const C& c) noexcept(noexcept(std::rbegin(c)))

-> decltype(std::rbegin(c));

Same mistake of using the member functions instead of the std:: variant in a few other places.

eisenwave · 2025-11-17T14:59:55Z

source/iterators.tex

 \indexlibrary{\idxcode{ssize(C\& c)}}%
 \begin{itemdecl}
-template<class C> constexpr auto ssize(const C& c)
+template<class C> constexpr auto ssize(const C& c) noexcept(noexcept(c.ssize()))


Suggested change

template<class C> constexpr auto ssize(const C& c) noexcept(noexcept(c.ssize()))

template<class C> constexpr auto ssize(const C& c) noexcept(noexcept(c.size()))

AlisdairM mentioned this pull request Nov 17, 2025

[iterator.synopsis] The <iterator> header should be // mostly freestanding #8545

Open

P3016R6 Resolve inconsistencies in begin/end for valarray and braced …

d21204a

…initializer lists

AlisdairM force-pushed the p3016r6_inconsistent_begin_end branch from e507db3 to d21204a Compare November 17, 2025 08:03

AlisdairM added 2 commits November 17, 2025 15:06

[support.initlist.range] Removed section in xrefdelta.tex

09fe942

AlisdairM force-pushed the p3016r6_inconsistent_begin_end branch from fda8120 to 09fe942 Compare November 17, 2025 08:27

eisenwave added this to the post-2025-11 milestone Nov 17, 2025

eisenwave suggested changes Nov 17, 2025

View reviewed changes

	\rSec2[valarray.range]{range access}
	\rSec2[valarray.range]{Range access}

		\indexlibrarymember{begin}{valarray}%
		\indexlibrarymember{valarray}{begin}%

		\indexlibrarymember{end}{valarray}%
		\indexlibrarymember{valarray}{end}%

		template<class C> constexpr auto crbegin(const C& c) noexcept(noexcept(c.crbegin()))
		-> decltype(std::rbegin(c));

	template<class C> constexpr auto ssize(const C& c) noexcept(noexcept(c.ssize()))
	template<class C> constexpr auto ssize(const C& c) noexcept(noexcept(c.size()))

P3016R6 Resolve inconsistencies in begin/end for valarray and braced initializer lists #8544

Are you sure you want to change the base?

P3016R6 Resolve inconsistencies in begin/end for valarray and braced initializer lists #8544

Conversation

AlisdairM commented Nov 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

eisenwave left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

AlisdairM commented Nov 17, 2025 •

edited

Loading