public inbox for gcc-bugs@sourceware.org help / color / mirror / Atom feed
From: "eteran at alum dot rit.edu" <gcc-bugzilla@gcc.gnu.org> To: gcc-bugs@gcc.gnu.org Subject: [Bug libstdc++/108645] New: Change in behavior, std::accumulate doesn't always work as expected in C++20 builds Date: Thu, 02 Feb 2023 21:00:51 +0000 [thread overview] Message-ID: <bug-108645-4@http.gcc.gnu.org/bugzilla/> (raw) https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108645 Bug ID: 108645 Summary: Change in behavior, std::accumulate doesn't always work as expected in C++20 builds Product: gcc Version: unknown Status: UNCONFIRMED Severity: normal Priority: P3 Component: libstdc++ Assignee: unassigned at gcc dot gnu.org Reporter: eteran at alum dot rit.edu Target Milestone: --- I encountered an interesting change in behavior today involving playing with std::move_iterator types using the following code: ``` #include <vector> #include <iostream> #include <string> #include <numeric> #include <iterator> void print_v(const char *rem, const std::vector<std::string> &v) { std::cout << rem; for (const std::string &s : v) std::cout << '"' << s << '"' << ' '; std::cout << '\n'; } int main() { std::vector<std::string> v = {"this", "_", "is", "_", "an", "_", "example"}; print_v("Old contents of the vector: ", v); std::string concat = std::accumulate(std::make_move_iterator(v.begin()), std::make_move_iterator(v.end()), std::string()); print_v("New contents of the vector: ", v); std::cout << "Concatenated as string: " << '"' << concat << '"' << '\n'; } ``` The expected output is: ``` Old contents of the vector: "this" "_" "is" "_" "an" "_" "example" New contents of the vector: "" "" "" "" "" "" "" Concatenated as string: "this_is_an_example" ``` And that is the output I get when compiling with `-std=c++17`. So the expectation is that the moved from strings become empty. However, when using C++20, where `std::accumulate` is now defined to use `std::move`, the output is this instead: ``` Old contents of the vector: "this" "_" "is" "_" "an" "_" "example" New contents of the vector: "this" "_" "is" "_" "an" "_" "example" Concatenated as string: "this_is_an_example" ``` A few thoughts: 1. I'm not sure this is a "bug" since moved from objects are in an unspecified, valid state. Being "unchanged" seems to fit that description just fine. 2. I'm guessing that this is a weird situation where the `operator+` overload that has a `basic_string &&` and the lhs isn't stealing the guts as expected for some reason. Tested with both clang and gcc trunk, so it seems to be a library level issue. C++20 Example: https://godbolt.org/z/7MMqWbxEr C++17 Example: https://godbolt.org/z/6YnrrE5zf
next reply other threads:[~2023-02-02 21:00 UTC|newest] Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top 2023-02-02 21:00 eteran at alum dot rit.edu [this message] 2023-02-02 21:05 ` [Bug libstdc++/108645] " eteran at alum dot rit.edu 2023-02-03 1:40 ` pinskia at gcc dot gnu.org 2023-02-03 9:32 ` redi at gcc dot gnu.org 2023-02-03 10:01 ` redi at gcc dot gnu.org 2023-02-03 10:08 ` redi at gcc dot gnu.org
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=bug-108645-4@http.gcc.gnu.org/bugzilla/ \ --to=gcc-bugzilla@gcc.gnu.org \ --cc=gcc-bugs@gcc.gnu.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).