- Boost - lists.boost.org

Compression libraries such as zlib and zstd ought to be first class boost libraries
by Nigel Stewart 05 May '24

05 May '24

Hello all, I've been having a fairly hard time enabling zlib and zstd for boost 1.85.0 so we can make use of the compression support in iostreams module. We do have some specifics - static linking, non-standard file names and not system-installed. Windows and Linux. For various reasons it looks like we're going to copy certain files into our modules, rather than puzzle over making this work. It's not ideal, but workable. But along the way it's become clear that this functionality is not first-class boost citizen. Such as the broken links in the documentation, and various historical questions via google reporting similar issues. There are also other copies of zlib in the beast tree, because it's obviously needed for HTTP stuff too. zstd too? So I did think that I should at least mention that zlib, zstd, bzip2 and xv-utils all ought to be promoted to first-class boost libraries even though the majority of the maintenance is coming from upstream. They ought to be part of the "big tarball" and enabled and built by default regardless of what happens to be installed on the system. The cmake stuff ought to "just work" the way it does for other boost modules. I'm not quite volunteering myself, but it seems like a start to pose the question. The developer experience could be significantly improved for this particular corner of things. - Nigel Stewart

7 24

Proposal for yet another MQTT library for Boost
by Takatoshi Kondo 01 May '24

01 May '24

Dear Boost Developers, I'd like to propose a MQTT library to the Boost Libraries. Repository link: https://github.com/redboltz/async_mqtt Document link: https://redboltz.github.io/async_mqtt/doc/5.0.0/index.html Here is characteristics of async_mqtt: - C++17 or later is required. - async_mqtt supports Boost.Asio style completion tokens. - e.g. use_future, deferred, use_awaitable, callback functions, etc - Not only MQTT client but also MQTT server is supported. - User can use this library to develop MQTT broker. - Sample MQTT broker application is included. - Both MQTT v3.1.1 and v5.0 are supported. - In addition, undetermined protocol version is supported for server implementation. In this case, the protocol version is decided when you receive CONNECT packet. - Continuous packet sending - Users can send the next MQTT packets before the previous sending is not completed (completion handler is not invoked yet). - Auto acquiring/mapping topic alias is supported. - In addition, when resending a PUBLISH packet after reconnecting, the topic alias is removed and the original topic is restored. It is required by MQTT v5.0 spec. - Packet based send()/recv() APIs. - MQTT has various packets and some of their fields are optional. In order to keep simple APIs, async_mqtt uses packet based APIs. - See https://redboltz.github.io/async_mqtt/doc/5.0.0/tutorial/send_recv.html - Layered archtecture - async_mqtt can have predefined TCP, TLS, WS, WSS, and user defined underlying layer. Users can access the underlying layer using next_layer() member function of MQTT endpoint. So you can configure any layers by yourself. e.g. tcp_no_delay, set send/recv buffer size, etc - See https://redboltz.github.io/async_mqtt/doc/5.0.0/tutorial/create_endpoint.ht… - High performance - See https://redboltz.github.io/async_mqtt/doc/5.0.0/performance.html Background I have developed another mqtt library mqtt_cpp https://github.com/redboltz/mqtt_cpp since 2015. It is widely used but has some limitations. For example: - The functions are directly mapped to packets(publish, subscribe, etc), so optional features are not designed cleanly. - Due to set_publish_hander / on_publish style handler design, Completion Token cannot be supported. Based on this 9years experience, I designed the new MQTT library. This is async_mqtt. I would appreciate if you could endorse this library. Regards, Takatoshi Kondo

3 4

Re: [boost] modern logic classes
by g.peterhoff＠t-online.de 29 Apr '24

29 Apr '24

Here is an update. Since a few boost libraries use tribool, these would have to be converted to bool3. Unfortunately it is not that easy to establish a new boost library. Especially if it is incompatible with previous implementations. thx Gero

1 0

Can we get rid of auto-linking?
by René Ferdinand Rivera Morell 29 Apr '24

29 Apr '24

Today I got bit by auto-linking again. As I forgot to define BOOST_ALL_NO_LIB in a CI build. I can't remember a time when auto-linking has been beneficial. And I've lost the number of times I've had to answer user questions with "Auto-linking is getting in the way. Define BOOST_ALL_NO_LIB and link normally." more times than I can count. If there was ever a benefit to auto-linking it has not surpassed the drawbacks. So.. Can we get rid of auto-linking? -- -- René Ferdinand Rivera Morell -- Don't Assume Anything -- No Supone Nada -- Robot Dreams - http://robot-dreams.net

7 15

Emitting complete type information in boost-python docstrings
by Jakob van Santen 29 Apr '24

29 Apr '24

Hello, I’m not sure if this is the right forum for this question, but here goes: I work on a large scientific codebase that is heavily tied to boost, and that most users interact with via boost-python. Lately I have been working on generating PEP 484 type stubs for the ~1000 classes it exposes in order to enable static type checking and code completion in editors. This required some modifications to boost-python itself in order to get complete type information into the generated docstrings, where it can be consumed by external tooling to generate stubs: https://github.com/boostorg/python/pull/417 I have found this quite useful, and suspect it might be useful for others as well to have it in a Boost release. Can anyone give me a pointer as to how to make that happen, or let me know where or whom I should be asking instead? I asked on cplusplus-sig(a)python.org <mailto:cplusplus-sig@python.org>, which was the place many years ago, but that list seems to have died out. Many thanks in advance! Cheers, Jakob ------------------------------------------------------- Jakob van Santen DESY - Deutsches Elektronen-Synchrotron Platanenallee 6, D-15738, Zeuthen Office 2L/32 +49 (0)33762-7-7233 -------------------------------------------------------

2 1

BOOST_THREAD_HAS_EINTR_BUG...
by David Bien 26 Apr '24

26 Apr '24

Not really sure where to bring this issue up. I have a singleton globals object with a session manager, share manager, etc. that is destroyed atexit() on final reference release. There are a few mutexes in the implementation. This functions fine on Windows, MacOS, etc - the globals object is released during atexit() and shutdown proceeds normally. Under Linux when BOOST_THREAD_HAS_EINTR_BUG is defined we SIGABRT at this: ~mutex() { BOOST_VERIFY(!posix::pthread_mutex_destroy(&m)); } The verify fails and we abort(). When I define BOOST_THREAD_HAS_NO_EINTR_BUG we do not abort - which means that posix::pthread_mutex_destroy() returns 0 in the case where BOOST_THREAD_HAS_NO_EINTR_BUG is defined and returns non-zero in the case of it not being defined. Guidance? Also point me at a better place for this discussion if there is such a place. Thanks, David Bien

3 10

question about singleton_pool
by 李重斌 26 Apr '24

26 Apr '24

I want to use singleton_pool to manage memory. This is just an example which tests it. My question: the first time i allocate 100 spot object, and free them. If i allocate some spot objects, and the number of them is less than 100. no more memory is needed, there is no need to allocate memory from os again. but the fact is another memory block is allocated. The first block is 4016 bytes (sizeof(Spot) = 40), the second block is 8016 bytes. This is not as my imagine. Is there something wrong with these code? ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// std::cout << "size of Spot " << sizeof(Spot) << " bytes\n"; std::cout << "test singleton_pool firt time\n"; const size_t NumSpots = 100; Spot* p1 = static_cast<Spot*>(spot_pool::ordered_malloc(NumSpots)); for (size_t i = 0; i < NumSpots; i++) { p1[i].x = 1; p1[i].y = 1; } spot_pool::ordered_free(p1); std::cout << "test singleton_pool second time\n"; Spot* p2 = static_cast<Spot*>(spot_pool::ordered_malloc(NumSpots/2)); for (size_t i = 0; i < NumSpots/2; i++) { p2[i].x = 2; p2[i].y = 2; } spot_pool::ordered_free(p2); /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////

1 0

UUID design discussion
by Peter Dimov 26 Apr '24

26 Apr '24

We've been discussing some UUID design issues in, well, issues in the Uuid repo, namely https://github.com/boostorg/uuid/issues/96 and, for context, https://github.com/boostorg/uuid/issues/113. This topic might be of wide interest to developers, so I thought it would be better to have this list know about it. Anyone who has an opinion on the matter is welcome to express it in the issue, or here. In addition to the questions raised in #96, namely, how should the constructors and the accessors of `uuid` look like (once we somehow move away from the public `data` member), here are some more topics of interest: * Should a default-constructed `uuid` remain uninitialized, or should the default constructor produce the nil Uuid? (Uninitialized is not a good look in 2024 because safety. On the other hand, `new uuid[ 1048576 ]` will start doing a `memset`.) * At the moment wide strings are processed by the name generators by converting every wchar_t to 32 bit, then hashing the bytes, zeroes and all. This doesn't strike me as correct. I think that the string should be converted to UTF-8 on the fly (with 32 bit wchar_t assumed UTF-16 and 32 bit wchar_t assumed UTF-32.)

5 11

Re: [boost] modern logic classes
by g.peterhoff＠t-online.de 25 Apr '24

25 Apr '24

>> Hello, >> I would like to provide modern logic classes, please see https://github.com/boostorg/logic/issues/30 >> What are your opinions/comments on this? > Interesting idea. > But rather than messing with tribool, I'd prefer to just see a new library quadbool. I would also like tribool to be removed. But until then it still needs to be supported. regards Gero

1 0

[stacktrace] boost_stacktrace_from_exception on non-x86
by Tim Haines 24 Apr '24

24 Apr '24

This feature is guarded by BOOST_STACKTRACE_ALWAYS_STORE_IN_PADDING, but that is only enabled on x86 ( https://github.com/boostorg/stacktrace/blob/a33e198cf660c213e700d6f050a4641…) This means that ./bootstrap.sh --with-toolset=gcc --with-libraries=stacktrace ./b2 install threading=multi toolset=gcc always fails to build on non-x86 platforms. There are three current workarounds: 1) set boost.stacktrace.from_exception=off 2) set BOOST_STACKTRACE_LIBCXX_RUNTIME_MAY_CAUSE_MEMORY_LEAK 3) build with threading=single (1) prevents me from using this feature. (2) shouldn't be needed on it's face because I'm using gcc/libstdc++. (3) affects more than stacktrace. Is this an incidental error, or is there a reason I wasn't able to deduce as to why this feature can't be used on non-x86+libstdc++? It seems like the check for libc++ on non-x86 can't be entirely done at compile time. I definitely understand wanting to alert users to the dangers of using this feature, but there also doesn't seem to be a good way of detecting the presence of libc++ at compile time (hence the need for https://github.com/boostorg/stacktrace/blob/a33e198cf660c213e700d6f050a4641… ). Thanks. - Tim

1 0