- Boost-users - lists.boost.org

[Transitive Dependencies] Assessing Secondary Dependency Impact on Primary Modules
by joseph.curran＠gehealthcare.com 29 Jul '25

29 Jul '25

Hello, We’re currently evaluating the impact of continuing to use an older version of Boost by reviewing fixes and changes introduced in later versions. While our application has several direct dependencies, the bcp tool reveals that we also pull in dozens of indirect (transitive) dependencies. Ideally, we’d like to limit our review to the release notes and fixes for the direct dependencies—focusing on how our application uses those modules directly. However, we’re concerned about missing issues in secondary dependencies that could affect the behavior of the primary modules we depend on. For example: Our application uses moduleA, which internally includes moduleB. If moduleB had a memory leak that was fixed in a later release, how can we determine whether moduleA was impacted by that issue - and whether our application was indirectly affected? Questions: Is there an efficient way to identify whether such secondary dependency issues have a downstream impact on the primary modules? That is, without deep diving into both moduleA and moduleB code bases and seeing if the issues/fixes apply. Are there any Boost policies or guidelines stating whether maintainers of a primary module are expected to review and address issues in their transitive dependencies? Any insights or suggestions would be greatly appreciated. Thanks, Joe

1 0

Polygon - Voronoi - clip to rectangular bounding box
by neil.j.chisholm＠gmail.com 20 Jul '25

20 Jul '25

Hello I am a hobby programmer and I hope to make use of the polygon library to procedurally generate voronoi diagrams for the purpose of defining world maps in a videogame. My use case includes a requirement that the voronoi diagram be clipped to a rectangular bounding box. The corollary to this requirement is that every voronoi cell is therefore (as a result of being clipped to the bounding box where necessary) a complete convex polygon. By this I mean that an edge must not extend outside the bounding box. If the edge would extend beyond the bounding box, it must be be clipped to the bounding box creating a new vertex at the intersection of the edge and the bounding box. I believe there would also need to be a vertex created at each corner of the bounding box. In order for a voronoi cell to be a complete convex polygon, new edge(s) must be created which run between the newly created vertices. My understanding is that the polygon library does not include a method to clip edges to a bounding box. Do you have any recommendations about how to approach defining a method which does so? - It should be possible to define the rectangular bounding box dimensions similarly to how points and segments are defined - For the sake of efficiency I assume it would be appropriate only to consider edges which extend beyond the given bounding box - While a general solution should probably be able to clip parabolic edges, my use case will likely only involve sites comprised of points (no segments) so a solution clipping straight line edges only would be sufficient for my own purposes If you have any thoughts I would be very grateful to receive them. Many thanks in advance.

1 0

[release] Boost 1.89.0 beta 1 released
by Marshall Clow 17 Jul '25

17 Jul '25

Boost release 1.88.0 beta 1 is now available at: https://archives.boost.io/beta/1.89.0.beta1/source/ The SHA256 checksums are: 9ece9af2b243f90b76f679b278b512b6589c7843e4dd54cffcb719bbc0801998 boost_1_89_0_b1.tar.bz2 bf45af4a58f39c2ec374ef6e70b26859b786ba2ed087bc869d92a2b4de14e0d2 boost_1_89_0_b1.zip f255b4a31558b3fb81be1dc7e8b367fb9997dbe5e71da98a730c681377abffa9 boost_1_89_0_b1.7z 984ee193ddf192795ba635d4c73b2b43e1e596995d7ce5519a3a72fab9bc0b40 boost_1_89_0_b1_rc1.tar.gz Release notes (still need updates): https://original.boost.org/users/history/version_1_89_0.html Please try the beta and report any problems you encounter. Thanks to everyone who contributed to the release. -- Boost Release Team

1 0

lowest_layer() is deprecated in boost::beast, replaced by boost::beast::get_lowest_layer(), but still in use by boost::asio:ssl::stream
by Klebsch, Mario 14 Jul '25

14 Jul '25

Hi, The boost libraries beast and asio seem to disagree on whether a socket/stream should offer the lowest_layer() method or not. Boost::beast::tcp_stream has only private versions of this method and the associated typedef lowest_layer_type, while e.g. boost::asio::ssl::stream still requires this method. The replacement get_lowest_layer() is only offered in beast. For boost::asio::ssl::stream this is handled by boost::beast::tcp_stream declaring boost::asio::ssl::stream as friend, so offering access to lowest_layer(). But I have my own ssl stream, based on a different SSL library than openssl, and this stream class is also (because it is based on boost::asio::ssl::stream) dependant on lowest_layer(). I think, boost::beast and boost::asio should agree on whether to deprecate lowest_layer() and offer get_lowest_layer(), or not. Klebsch Mario Funktion | R&D Tel: +49 (0) 531 38 701 718 Raum: Braunschweig, E20 Diese E-Mail und die an sie angehängten Dateien sind ausschließlich für Personen oder Institutionen bestimmt, deren Namen oben aufgeführt sind. Sie können Informationen enthalten, die durch das Berufsgeheimnis geschützt sind und deren Weitergabe strengstens untersagt ist. Jede elektronische Nachricht kann geändert werden. ACTIA lehnt jede Verantwortung für diese Nachricht ab. Der Inhalt dieser Nachricht stellt keine Verpflichtung seitens unseres Unternehmens dar. Wenn Sie kein Empfänger sind, weisen wir Sie darauf hin, dass das Lesen, Vervielfältigen oder Verteilen strengstens untersagt ist. Wir bitten Sie daher, uns umgehend über diesen Brief zu informieren und diese Nachricht sowie eventuell beigefügte Unterlagen aus Ihrem Postfach zu löschen. Danke. This e-mail and the files attached to it are intended exclusively for persons or institutions whose names are listed above. They may contain information that is protected by professional secrecy and the disclosure of which is strictly prohibited. Any electronic message can be modified. ACTIA declines all responsibility for this message. The content of this message does not represent a commitment on the part of our company. If you are not a recipient, we would like to point out that reading, copying or distribution is strictly prohibited. We therefore ask you to inform us immediately about this letter and to delete this message and any accompanying documents from your mailbox. Thank you.

1 0

Free threaded python support with PEP 779
by george.sykes＠connectorsubsea.com 09 Jul '25

09 Jul '25

Hi, With Python 3.14 introducing free threaded python I was wondering if anyone's had a chance to try this out with Boost Python? Seems like it should be able to help us increase performance on certain types of tasks but not sure if Boost Python supports using the functionality. If not, i'll go give it a crack and see what it's like. Best wishes George

1 0

Issue with Python 3.12 and Concurrent Threads
by jholbrook＠kinetic-vision.com 19 Jun '25

19 Jun '25

Hi, I have a Windows application that has been using Boost Python for many years now, and we are looking to upgrade from Python 3.7 to Python 3.12. This is a C++ application written in Qt5 that uses Boost Python to interface with Python scripts that do some scientific calculations, and the results are displayed in the C++ application. As such, these calculations are by and large done on a concurrent thread so as not to block the UI; however, some quick python functions are run from the main thread. The problem is that *something* changed in the Python 3.12 upgrade, and I'm not sure what -- but now it crashes when I run Python code from a different thread from the one I called Py_Initialize() on. I've tested up to Python 3.11, and I can Py_Initialize() on the main thread and call Python code from any thread and everything works. But starting in Python 3.12, that's no longer the case. If I initialize Python on the UI thread, I can only run it from that thread. If I initialize it on a concurrent thread, I can only run it on a concurrent thread. Since I am using Python, Qt5, and Boost Python, I am not 100% sure where the fault lies, but I am posting this here in case anyone has run into this issue. I haven't found any other references to it online. I can post a minimal example project if that would be helpful for debugging. Essentially, this is the meat of my code: //Initialization Function: Py_Initialize(); //Boost Python Code: boost::python::object main = boost::python::import("__main__"); boost::python::object name_space(main.attr("__dict__")); boost::python::exec_file("testPython.py", name_space, name_space); double testNumber = boost::python::extract<double>(name_space["testNumber"]); qDebug() << testNumber; //print(testNumber) #testPython.py: import random testNumber = random.random() Thanks for any help you can provide.

1 1

Issue with Python 3.12 and Concurrent Threads
by jholbrook＠kinetic-vision.com 18 Jun '25

18 Jun '25

Hi, I have a Windows application that has been using Boost Python for many years now, and we are looking to upgrade from Python 3.7 to Python 3.12. This is a C++ application written in Qt5 that uses Boost Python to interface with Python scripts that do some scientific calculations, and the results are displayed in the C++ application. As such, these calculations are by and large done on a concurrent thread so as not to block the UI; however, some quick python functions are run from the main thread. The problem is that *something* changed in the Python 3.12 upgrade, and I'm not sure what -- but now it crashes when I run Python code from a different thread from the one I called Py_Initialize() on. I've tested up to Python 3.11, and I can Py_Initialize() on the main thread and call Python code from any thread and everything works. But starting in Python 3.12, that's no longer the case. If I initialize Python on the UI thread, I can only run it from that thread. If I initialize it on a concurrent thread, I can only run it on a concurrent thread. Since I am using Python, Qt5, and Boost Python, I am not 100% sure where the fault lies, but I am posting this here in case anyone has run into this issue. I haven't found any other references to it online. I can post a minimal example project if that would be helpful for debugging. Essentially, this is the meat of my code: //Initialization Function: Py_Initialize(); //Boost Python Code: boost::python::object main = boost::python::import("__main__"); boost::python::object name_space(main.attr("__dict__")); boost::python::exec_file("testPython.py", name_space, name_space); double testNumber = boost::python::extract<double>(name_space["testNumber"]); qDebug() << testNumber; //print(testNumber) #testPython.py: import random testNumber = random.random() Thanks for any help you can provide.

1 0

Assignment of boost::dll::import_symbol to boost::shared_ptr fails to compile
by Yuri 14 Jun '25

14 Jun '25

I've encountered the compilation error with this program with boost-libs-1.88.0: #include <boost/dll/import.hpp> #include <boost/shared_ptr.hpp> class API { public: API() { } }; void x() { boost::shared_ptr<API> pluginInterface; pluginInterface = boost::dll::import_symbol<API>("path", "name"); } The error: $ c++ -std=c++20 -c s.cpp -I /usr/local/include/ s.cpp:13:21: error: no viable overloaded '=' 13 | pluginInterface = boost::dll::import_symbol<API>("path", "name"); | ~~~~~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ /usr/local/include/boost/smart_ptr/shared_ptr.hpp:407:18: note: candidate function not viable: no known conversion from 'boost::dll::detail::import_type<API>' (aka 'st d::shared_ptr<API>') to 'const shared_ptr<API>' for 1st argument 407 | shared_ptr & operator=( shared_ptr const & r ) noexcept | ^ ~~~~~~~~~~~~~~~~~~~~ Compiler: clang-19, but gcc-12 fails in a similar way. It seems that such assignment should work. This error was originally found in the Dakota project (https://github.com/snl-dakota/dakota) This error prevents the Dakota project from compiling. Thanks, Yuri

2 1

mailman upgrade
by Sam Darwin 04 Jun '25

04 Jun '25

Hi, As discussed earlier on the developer list, mailman2 is end-of-life and no longer supported. We are planning to upgrade the mailing lists from mailman2 to mailman3. This will be done in-place, importing the previous archives and user list into a mm3 instance hosted at the same URL lists.boost.org. An example staging instance can be viewed at https://lists.preview.boost.org Mailman3 is more complex than Mailman2 with multiple components including "web" and "core". After the upgrade your account should still exist in the core backend, which means you will continue to receive messages and be able to send emails. However, to interact with the web frontend (optional) you must register an account using the same email address at https://lists.boost.org. Registering in the web portal will require setting up a new password, and confirming the email address. Be sure to use your existing email from before. Any previous passwords will not be valid. If you don't create an account in the web UI, your email subscriptions to the lists continue. But you can't yet log in to configure the account. No action is needed right now. Wait until after the upgrade. All previous messages will be imported to the new system. Additionally, the earlier archives in the form of static plain html pages will be available at https://listarchives.boost.org/ The imported lists are Boost, Boost Announce, and Boost Users. Activity on any other lists is remarkably low such as 1-2 emails per year. Please migrate other discussions to Github Issues, Slack, or one of the main mailing lists. Let us know any concerns. Report issues in the github repository cppalliance/boost-mailman. The tentative schedule is Tuesday June 17. Plan on not using the list for one day, or perhaps a few days. Roll back to mailman2 if something unusual is discovered soon after going live. Chat out-of-band in cpplang slack boost channel.

1 0

boost::process v1 broken in 1_88_0
by Sebastian Koerner 02 Jun '25

02 Jun '25

Hi Boost-Users, I understand that boost process has v2 and v1 is deprecated, but I thought v1 will at least work on for existing stuff. Between 1_87 and 1_88 process v1 stopped working with v1 code. Something changed v1-incompatible with the includes, like: boost::process::child is unknown. error: ‘child’ is not a member of ‘boost::process’ 73 | std::vector<boost::process::child> m_processes; Is this known or easy-to fix?

2 2