- Boost - lists.preview.boost.org

[charconv] Review Results and Report
by Christopher Kormanyos 02 Feb '24

02 Feb '24

Dear Boost Community, The review of the proposed Boost.Charconv is finished It took place 15-25 January 2024. REVIEW RESULT ----------------------------- The result is Boost.Charconv is unconditionally * ACCEPTED * into Boost. Thank you Matt. Boost.Charconv will enrich Boost and surely be used vigorously. REVIEW REPORT ----------------------------- The review tally is: * 1 Conditional ACCEPT * 4 ACCEPT * 1 Positive Document Review ---------------------------- = 5 ACCEPT, since the single condition has already been satisfied in develop branch. My Comments and Descriptions below are based on limited knowledge and my subjective interpretations. Please correct, add, modify these (and to these) as needed. * Comments: The review atmosphere was highly productive and excellent. Reviews and contributions featured deep technical content and detail. Fact-based statements combined with experienced-based opinions were provided. There was great strength of expression and clarity. Yet there was mutual respect and fairness, even in the course of lively debate. * Descriptions: Ultimately there were three areas receiving the most debate. For background, see references below. 1) The library needs to allocate. Among other conversions, Boost.Charconv handles conversion of floating-point values to chars. Edge-cases may require hundreds or even thousands of bits. * Consensus: Handle the required memory needed, as done, with low-level C-LIB malloc. But do describe this decision with greater depth of clarity in the docs. 2) The project makes use of a variety of methods, including some table-driven methods. One example is RYU (see [1]). A quick glance at Fig. 4 in [1], for instance, reveals the need for relatively large, constant-valued tables. Largescale tables, combined with the C++11 design goal formed a clear indication to produce a compiled library (as opposed to a header-only one). * Consensus: Compiled-library is sound and viable. Tables (especially in constexpr-context) might bloat or not even work in some language-standards environments. Some commenters, however, seemed subjectively to disapprove of this consensus, feeling that header-only offers the best chances at friction-free setup and convenient portability. Other commenters revealed a sense that compiled libraries are gaining traction in the industry. They feel that this is because the community, their available tools, language-skills, specifications, compilers, CI, VMs, etc. have matured significantly and continue to mature. Thereby they embrace a perceived future-oriented trend that views client-builds to be uncomplicated. There were no blockers on the consensus. 3) The top-level interface differs from that of std-<charconv> in points such as overloads and error-handling. Whereas some details are not even fully hashed-out in the Standard. * Consensus: Weigh known opinions and decide interface details based on unbiased majority. My current understanding is: Offer two drop-ins for from_chars, with the caveat that one should provide better error-handling. REVIEW ACKNOWLEDGMENTS ----------------------------- Thanks again, Matt for Boost.Charconv. Many thanks to our expert reviewers, commenters and watchers. Your contributions keep Boost excellent! Kindest regards, Christopher Kormanyos Boost.Charconv Review Manager * References: [1] Ulf Adams, "RYU: fast float-to-string conversion", ACM SIGPLAN Notices, Volume 53, Issue 4, pp 270–282 (ACM 2018) [2] "Number Parsing at a Gigabyte per Second", Daniel Lemire's Blog, https://lemire.me/blog/2021/01/29/number-parsing-at-a-gigabyte-per-second/ [3] https://github.com/lemire/fast_double_parser

1 0

Using ASIO to wait on a foreign socket being readable, with a timeout
by Dominique Devienne 31 Jan '24

31 Jan '24

Hi, I'm trying to use ASIO to wait until a socket is ready to read, or timeout. This is a blocking wait. But the socket is not "owned" by ASIO, but is from the libpq PostgreSQL connection. So ASIO should NOT consume any of the input on the socket, just notify me when that socket is "read-ready". I've cobbled together the following test code: ``` const int socket2 = PQsocket(conn2.handle()); BOOST_CHECK(socket2 > 0); int timr_err_code = 0; int sock_err_code = 0; bool timr_aborted = false; bool sock_aborted = false; std::string timr_err_msg; std::string sock_err_msg; bool timed_out = true; auto c1 = "C# 1"sv; conn2.listen(c1); conn1->notify(c1, "wake up!"sv); { boost::asio::io_context io_ctx; boost::asio::ip::tcp::socket io_sock(io_ctx); io_sock.assign(boost::asio::ip::tcp::v4(), socket2); boost::asio::steady_timer io_timer(io_ctx); io_timer.expires_from_now(std::chrono::milliseconds(100)); io_sock.async_wait( io_sock.wait_read, [&](const boost::system::error_code& errc) { if (errc) { sock_err_code = errc.value(); sock_err_msg = errc.message(); sock_aborted = (errc == boost::asio::error::operation_aborted); } timed_out = false; //io_timer.cancel(); io_ctx.stop(); } ); io_timer.async_wait( [&](const boost::system::error_code& errc) { if (errc) { timr_err_code = errc.value(); timr_err_msg = errc.message(); timr_aborted = (errc == boost::asio::error::operation_aborted); } io_sock.cancel(); io_ctx.stop(); } ); io_ctx.run(); // blocks until either async op above runs } // there's a notif to read, so io_sock.async_wait should have completed BOOST_CHECK_EQUAL(timr_err_code, 0); BOOST_CHECK_EQUAL(sock_err_code, 0); BOOST_CHECK_EQUAL(timr_err_msg, ""); BOOST_CHECK_EQUAL(sock_err_msg, ""); BOOST_CHECK_EQUAL(timr_aborted, false); BOOST_CHECK_EQUAL(sock_aborted, false); BOOST_CHECK_EQUAL(timed_out, false); auto notif = conn2.notification(); BOOST_REQUIRE(notif); BOOST_CHECK_EQUAL(notif.backend_pid(), pid1); BOOST_CHECK_EQUAL(notif.channel(), c1); BOOST_CHECK_EQUAL(notif.payload(), "wake up!"sv); ``` The io_sock.async_wait(io_sock.wait_read, ...) succeeds, the IO loop stops as expected, but when I try to PQconsumeInput() from the LIBPQ connection next (as shown in the code below) ``` Notification Connection::notification() { auto [lock, hndl] = ensure_handle(); if (1 != PQconsumeInput(hndl)) { throw std::runtime_error(error_msg()); // <<<<<< THROWS } PGnotify_uptr notif{ PQnotifies(hndl) }; return Notification(std::move(notif)); // may be null } ``` that calls fail with this error message: fatal error: in "NotificationTests/test_across_backends": class std::runtime_error: could not receive data from server: Socket operation on non-socket (0x00002736/10038) LIBPQ didn't like what ASIO somehow did on the socket descriptor. The LIBPQ doc says to select() the descriptor, but I thought ASIO would be cleaner, more C++, and perhaps more cross-platform. Can somehow help, telling me if I'm using ASIO incorrectly here? Thanks, --DD

3 2

Boost.Charconv review runs January 15th - 25th, 2024
by Christopher Kormanyos 31 Jan '24

31 Jan '24

Dear all, The review of Boost.Charconv by Matt Borland runs Monday, January 15th through January 25th, 2024. Code: https://github.com/cppalliance/charconvDocs: https://master.charconv.cpp.al/Review Schedule: https://www.boost.org/community/review_schedule.html Thank you for considering this review. Please optionally provide feedback on the followinggeneral topics: - What is your evaluation of the design? - What is your evaluation of the implementation? - What is your evaluation of the documentation? - What is your evaluation of the potential usefulness of the library? - Did you try to use the library? With what compiler? Did you have any problems? - How much effort did you put into your evaluation? A glance? A quick reading? In-depth study? - Are you knowledgeable about the problem domain? Christopher KormanyosBoost.Charconv Review Manager

15 76

[charconv] is std::from_chars practical?
by Andrzej Krzemienski 26 Jan '24

26 Jan '24

Hi Everyone, This is about https://github.com/cppalliance/charconv/issues/110, again. Now I understand Peter's point better. std::from_chars is impractical, or one would say "broken", at least for the case of ERANGE. Actually applying a patch in https://cplusplus.github.io/LWG/lwg-active.html#3081 will not make the tool good, it will just make it broken in another way. In that case, maybe the goal of providing literally a *drop-in replacement* for std::from_chars is not a good one? Maybe this is the case similar to variant2, where the community would benefit more from the library motivated by "how would from_chars look like if we designed it instead of WG21". That is, an *alternative to* std::from_chars, rather than a *drop-in-replacement for* std::from_chars. Regards, &rzej;

6 6

Review of proposed CharConv
by Antony Polukhin 25 Jan '24

25 Jan '24

## What is your evaluation of the design? Design follows the C++ Standard library design with all the latest additions. Everything is fine. ## What is your evaluation of the implementation? Fine in general. A few nitpicks/concerns follow. ### Arrays of constants RYU/fast_float/... are used for some of the conversions and most of those algorithms have huge tables full of constants. For example https://github.com/cppalliance/charconv/blob/41a83261c0f65aee575e89ccb71d94… and https://github.com/cppalliance/charconv/blob/41a83261c0f65aee575e89ccb71d94… and https://github.com/cppalliance/charconv/blob/41a83261c0f65aee575e89ccb71d94… Those constants have internal linkage. They are put in every translation unit that includes and uses them. Those constants would not be merged by linker. Moreover, the same constants with different names now exist in Standard Library implementations and are not merged with Boost.Charconv. To reduce the binary size and improve data locality I'd recommend switching to std implementation in sources if the feature test macro tells that std::to_chars/std::from_chars are fine. ### Internal headers in include/ Some of the headers are used only internally during build. It seems reasonable to move them to the src/ directory to make sure that no one uses them and that the library itself does not include them by accident. ### Header only support There is a PR to make the library header only https://github.com/cppalliance/charconv/pull/138 This feature is demanded by users however it raises the constants question again. Marking arrays of constants with `inline` should be done to make the constants externally visible, so that the linker could merge them. ### Constexpr for floats Just a feature request - it would be nice to have constexpr from_chars and constexpr to_chars for floating point types. If is_constant_evaluated() - do slow but constexpr conversion. Otherwise - fallback to std:: version or to the implementation in src/ ### Locales Locales are used in some corner cases along with switching them back and forth https://github.com/cppalliance/charconv/blob/41a83261c0f65aee575e89ccb71d94… Looks like there's no need in that line for locales switching. Just get the decimal point separator from the current locale and if it is not '.' then do the replacement of the first '.' to the locale's separator. ### Nitpicks Looks like zero initialization of the buffers could be skipped in following places: * https://github.com/cppalliance/charconv/blob/41a83261c0f65aee575e89ccb71d94… * https://github.com/cppalliance/charconv/blob/41a83261c0f65aee575e89ccb71d94… * https://github.com/cppalliance/charconv/blob/41a83261c0f65aee575e89ccb71d94… ## What is your evaluation of the documentation? Looks like parts of the docs repeat themself. For example, the reference section duplicates some of the previous descriptions. The latter could be just removed. Another thing that catches an eye are the benchmarks. Please also provide some absolute numbers because "X is five times faster than Y" could easily be that X and Y execute for ~1ns each. Those numbers are profitable for users. Side Note for myself: Seems that LexicalCast shows poor performance in some cases where it should not. Will investigate. ## What is your evaluation of the potential usefulness of the library? Very useful for libraries that wish to support pre-C++17 standards or a wide range of compilers. Useful for experimenting and prototyping new C++ Standard Library proposals ## Did you try to use the library? With what compiler? Did you have any problems? ## How much effort did you put into your evaluation? A glance? A quick reading? In-depth study? Just looked through the code and the docs. ## Are you knowledgeable about the problem domain? I'm maintaining the Boost.LexicalCast and implemented a bunch of optimizations for std::to_chars/std::from_chars for the libstdc++. I'm not knowledgeable of RUY internals, but have the general understanding of the <charconv>. ## Do you think the library should be accepted as a Boost library? Yes

3 2

Zach Laine CharConv review
by Zach Laine 25 Jan '24

25 Jan '24

- What is your evaluation of the design? It follows the design of the equivalent facility in std::, so this feels moot. No complaints with that part. My only complaint is that the semantics of boost::charconv::from_chars and boost::charconv::from_chars_strict seem backward to me. I know this was mentioned by other reviewers. I think that with near-identity to the standard version, having the same (or corresponding name within boost::) should imply the same semantics. This is classic adherence to The Principle of Least Surprise. I would much prefer that the varying version have the different name, and that the one with the standard's semantics should be from_chars. I also encourage the authors to track any changes to std::from_chars with boost::charconv::from_chars. - What is your evaluation of the implementation? I mean, there's a lot of it. I'm sure it's all necessary, but I only looked at about 10% of it. What I saw seemed reasonable and well organized. - What is your evaluation of the documentation? It's great. The library is smallish, and the docs are correspondingly bite-sized, with plenty of examples. I might add a little more front matter that explains why someone should use this library. The docs get there, but putting it in the first paragraph is better. Busy people will blow right past everything that isn't in the first paragraph or two. - What is your evaluation of the potential usefulness of the library? High. I think the fact that this is a C++17 feature that we still can't reliably use says it all. The fact that it's about the same as the std:: implementation, or sometimes a little faster, is even better. - Did you try to use the library? With what compiler? Did you have any problems? Yes. I tried it with GCC 12, in a small test program just to see it go. No issues. - How much effort did you put into your evaluation? A glance? A quick reading? In-depth study? I spent about 3 hours, mostly looking at code. I also read all the docs and gave it a quick try. - Are you knowledgeable about the problem domain? Not particularly. I have written lots and lots of parsers, but never a simple low-level one like those in charconv. Zach

3 3

test message 19
by Sam Darwin 24 Jan '24

24 Jan '24

content test message 19

1 0

test message 7
by Sam Darwin 18 Jan '24

18 Jan '24

contents of test message 7 ___

1 0

Review of Parser library scheduled
by Mateusz Loskot 17 Jan '24

17 Jan '24

Folk, Review of Zach Laine's Praser library has been scheduled [1] for February 19, 2024 - February 28, 2024 Big thanks to Marshall Clow for volunteering to be the review manager. [1] https://www.boost.org/community/review_schedule.html Best regards, -- Mateusz Loskot on behalf of Boost Review Wizards

1 0

Making Boost modular. IMPORTANT! URGENT! <insert-superlative-here>!
by René Ferdinand Rivera Morell 17 Jan '24

17 Jan '24

Dear fellow Boost authors, As some of you may have heard.. Some of us have been working on making it possible to fully consume Boost in a modular arrangement. First some FAQs about this: Q: What is a "modular arrangement"? A: It's when the libraries can be used, and hence built, without creating the monolithic headers, without needing the root build files, and without needing the libraries to be arranged in the usual root/libs/<name>. Q: Will a modular Boost remove the Boost release? A: No. The collection of libraries is still a single release. See recent discussions about this. Q: Will this change the testing? A: No, unless you want to. You will still be able to test the same non-modular way. But you could also test the modular way. Q: Will this require that we change the current Boost source structure? A: Yes. Unfortunately there is one restriction that adhering to a modular Boost requires. We would not allow sublibs. That is, that we can't support having libraries in root/libs/thing/<library> style locations. All libraries must be single libraries under the root/libs directory. Thankfully there's only a handful of such libraries. The root/libs/numeric/* group of libraries. Q: Why do we want a modular Boost? A: I'm not going to enumerate the reasons here. But let me just say.. That we've wanted a way to reduce, i.e. modularize, Boost since at least 19 years ago. When we added the "--without-<library>" build option to the root build file <https://github.com/boostorg/boost/blame/master/Jamroot#L107>. Now the work part.. Over the past year or so I've been making patches to the Boost build files, in my own repo, that cleans up the many, many, problems that make libraries depend on the current source structure. I've also made changes to B2 to make it easier/possible/automatic to deal with the new modular freedom. The B2 changes were naturally easy to put in since I control the repo and distribution. And the last bit of them shipped with the 1.83 release last Summer. But now I need to start pushing the changes to the many Boost repos to get this finished. A couple of weeks ago I started posting PRs to incrementally, and unobtrusively, make the changes to the Jamfiles. So far I have these: <https://github.com/boostorg/ublas/pull/189> <https://github.com/boostorg/interval/pull/38> I'm not sure how else to say this so.. PLEASE, PLEASE, merge those and future PRs ASAP. -- -- René Ferdinand Rivera Morell -- Don't Assume Anything -- No Supone Nada -- Robot Dreams - http://robot-dreams.net

7 12