New issue from Jonathan: Should flat adaptors use insert_range when available?

Author: jwakely

xml/issue4350.xml

@@ -0,0 +1,38 @@
+<?xml version='1.0' encoding='utf-8' standalone='no'?>
+<!DOCTYPE issue SYSTEM "lwg-issue.dtd">
+
+<issue num="4350" status="New">
+<title>Should flat adaptors use `insert_range` when available?</title>
+<section>
+<sref ref="[flat.map]"/>
+<sref ref="[flat.multimap]"/>
+<sref ref="[flat.set]"/>
+<sref ref="[flat.multiset]"/>
+</section>
+<submitter>Jonathan Wakely</submitter>
+<date>05 Sep 2025</date>
+<priority>99</priority>
+
+<discussion>
+<p>
+As specified in <sref ref="[queue.mod]"/>,
+`std::queue::push_range` is guaranteed to use `c.append_range` if that exists.
+For `flat_map` and its flat friends, we only ever use `c.insert` and never try
+to use `c.insert_range`, `c.append_range`, etc.
+LWG thinks the "as if by" wording allows implementations to use `insert_range`
+when the container type being adapted is `std::vector` but not for user-defined
+containers.
+Should the flat adaptors follow `std::queue` and guarantee that they will use
+the `xxx_range` member if they exist?
+That would mean that user containers need to ensure that any member functions
+with those names must behave as expected, but we already require that for
+`std::queue`.
+</p>
+</discussion>
+
+<resolution>
+<p>
+</p>
+</resolution>
+
+</issue>