August 2022 - Boost - lists.preview.boost.org

Library proposal - indexed_array
by Julien Blanc 12 Oct '23

12 Oct '23

Hi, I've written and released a (rather small) library which allows defining arrays with arbitrary index mechanisms (hence the name, indexed_array). This especially allows indexing with scoped enums, use non zero-starting indexes, or use sparse indexes. This is not a view, but an array in the sense that it owns the data, and provides the same guarantees and functionalities as std::array, such as triviality propagation, being able to reside in read-only program sections, etc. It also supports multidimensional indexing, and provides an indexed_span template class to access lower dimension content (which can be used outside indexed_array). Finally, it provides an opt-in safer way to initialize the array content, which is especially useful when using complex and/or changing indexing schemes (like based on externally generated files). The library is available here: https://github.com/Julien-Blanc-tgcm/indexed_array And the doc is here: https://julien-blanc-tgcm.github.io/indexed_array/ It requires C++17 and mp11, and has first-class support for describe-d enums. If there is some interest, i'd like to propose it for inclusion in boost. So i'm looking for people who would endorse it. Thanks and best regards, Julien

8 24

Boost.Text is looking for a (mini-)review manager
by Zach Laine 07 Dec '22

07 Dec '22

I've made all the recommended changes to Boost.Text that came up in the review. Glen, the previous review manager, recommended a mini review of Boost.Text once I'd made those changes. If you'd like to manage the mini-review, please let me know. For everyone, including potential reviewers and review managers, here are the major changes: * All the specific review feedback issues have been addressed. * The string layer is now gone. * The text layer got a major re-work. It is now a set of template-based adaptors (basic_text<>, vs. a text typedef. etc.), which lets everyone get what they want. You can change the underlying storage type (including whatever allocator you prefer), select UTF-8 or UTF-16, and change the normalization form. Glen recommended I drop this layer, but the library is way less useful without this stuff, and these changes address all the concerns raised about the text layer in the review. The updated docs can be found here: https://tzlaine.github.io/text/doc/html/index.html Oh yeah, and I added concept constraints to the whole library when you build in C++20 mode. Zach

17 46

[Units] Incorrect definitions for torque, newton_meter
by Albert Dvornik 04 Sep '22

04 Sep '22

#include <iostream> #include "boost/units/systems/si.hpp" #include "boost/units/systems/si/io.hpp" using boost::units::quantity; using boost::units::si::torque; using boost::units::si::newton_meter; using boost::units::si::newton; using boost::units::si::meter; using boost::units::si::radian; int main() { // works fine: quantity<torque, double> tau0 = 0.0 * newton_meter; // SURPRISE: compilation fails because newton_meter is not the // same as newton*meter! // quantity<torque, double> tau1 = 0.0 * newton * meter; // Compilation succeeds even though the units on the RHS do not // represent torque but torsional stiffness! quantity<torque, double> tau2 = 0.0 * newton * meter / radian; // newton * meter should give us a quantity with units of Newton- // meters, i.e. m^2 kg s^-2. What this actually prints out is "0 J", // which is reasonable since dimensionally J is the same as N m. std::cout << boost::units::to_string(0.0 * newton * meter) << std::endl; // One might hope that newton_meter would be the same thing as // newton * meter. Instead, the result is "0 m^2 kg s^-2 rad^-1" std::cout << boost::units::to_string(0.0 * newton_meter) << std::endl; // The "torque" unit is also messed up in the same way. std::cout << boost::units::to_string(torque()) << std::endl; }

5 8

additional math-functions
by Gero Peterhoff 04 Sep '22

04 Sep '22

Hello, I would like to add several missing math functions in boost or specialize/overload existing ones for more types. These should be constexpr if possible. My problems with the concrete implementations are: - Which C++ standard to use C++11/14/17/20/23 ? - Should these be implemented in ccmath and math ? a) ccmath requires C++17 (eg "if constexpr"), so it is not possible to simply call the ccmath version in math, since math requires C++11. b) On the other hand, the math functions are already included in the standard (since C++11), so it wouldn't be necessary to continue to provide them (especially since they are not constexpr). Where and how exactly and with which C++ standard can I add the functions ? thx Gero

4 10

Password_protected .csv file
by Daggumati, Rajesh 31 Aug '22

31 Aug '22

Hi Team, We can create an .csv file using std::ofstream() function. But my project requirement demands to create a .csv file with password protected. Ofstream has open() function to create a file but how to create a file with password protected? Please help me on this. Regards, Rajesh D Confidential - Company Proprietary Please be advised that this email may contain confidential information. If you are not the intended recipient, please notify us by email by replying to the sender and delete this message. The sender disclaims that the content of this email constitutes an offer to enter into, or the acceptance of, any agreement; provided that the foregoing does not invalidate the binding effect of any digital or other electronic reproduction of a manual signature that is included in any attachment.

4 4

Boost warning philosophy
by Scott Johnson 30 Aug '22

30 Aug '22

Hello, We are migrating a legacy codebase (embedded system though a PC architecture) from Centos 7.x, using various boost 1.5x components and GCC 7.3, to Ubuntu 22.04 and GCC 11.2 (the compiler that ships with this version of Ubuntu). As our application needs things that don't come with the "stock" Boost version that Canonical packages for 22.04, specifically Python 2.x support, I have downloaded and configured Boost 1.80 for our application to use. Unfortunately, our code development standard is -Wall -Werror (turn on all warnings and treat them as errors), with exceptions to -Wall only allowed grudgingly. And this version of Boost seems to break things, with all sorts of warnings being thrown that we simply don't see on the older version, distro, and toolchain. One of the biggest offenders seems to be things like #if _MSC_VER when this preprocessor symbol is not defined--GCC likes to warn that the symbol is being "defaulted" to zero in this context. (Defining it as 0 causes all heck to break loose, as there is also plenty of #ifdef _MSC_VER in the code). I've also seen warnings for overloaded virtual functions being hidden (a warning I would prefer not to disable), null pointer dereferences (ditto), and local variables being shadowed (ditto) from various Boost components. This is on production builds with -O2 optimization level. What is, if anything, the Boost policy towards compiler diagnostics of this sort? Is there a set of warnings we should be prepared to ignore if we use Boost with a more modern (and pickier) compiler? Should we perhaps consider another version of GCC? An older, more mature version of Boost (we don't need any specific features in the current one)? Thanks, Scott -- engineer_scotty (no, not that one) If life gives you lemons, drink Hefeweizen

3 2

[filesystem] becoming worse than the standard?
by Vinnie Falco 29 Aug '22

29 Aug '22

I see this in the commit log for boost.filesystem: Deprecated path construction/assignment/appending from container types. Users are advised to use string types and iterators instead of containers to construct/assign/append to paths. In v4, the support for containers is removed. This can't be right now, can it? Because std::filesystem supports path construction from anything meeting the requirements of Source which is kind of range-like. Boost.URL and its downstream libraries HTTP-Proto, HTTP-IO, Websocket-Proto, and Websocket-IO are depending on the container-based interface to achieve reasonably decent syntax and semantics despite the limitations of the path API (which are the fault of the standard and not Boost). If std::filesystem::path supports container-based parameters to assign() and append() then why is boost::filesystem::path removing it? Thanks

11 45

Re: [boost] [url] Andrzej's notes
by Vinnie Falco 25 Aug '22

25 Aug '22

On Mon, Aug 22, 2022 at 11:48 PM Andrzej Krzemienski <akrzemi1(a)gmail.com> wrote: > Like in this example from the docs: > https://www.boost.org/doc/libs/1_80_0/libs/beast/doc/html/beast/quick_start… > > http::request<http::string_body> req{http::verb::get, target, version}; > req.set(http::field::host, host); > req.set(http::field::user_agent, BOOST_BEAST_VERSION_STRING); > > At no point do I have to see or provide a full URL. "target" is the URL there, it came from the command line. It is a relative-ref. >> * modifiers which take un-encoded inputs have a wide contract: all >> input strings are valid >> - however the url might need to reallocate memory to encode the result > > Does set_port() taking a string_view fall into that category? > Unlike other parts, port has special requirements on the string > contents that cannot be satisfied by pct-encoding. Yeah, we have a problem here. It sounds like we need to design an extra set functions. Maybe: url_base& url_base::set_port( string_view ); // throws result<void> url_base::try_set_port( string_view ); // returns result what do you think about that? > If you used result<>, you would lose the ability to chain the setters. Yeah... well, I think I'm OK with that. > As an alternative, you could say that function set_port() has a precondition: > the input string has to represent a number, the caller is responsible for that, > and set_port() performs no validation, and therefore throws nothing. But I > guess that would violate one of the design goals of the library: "securely, > including the case where the inputs come from untrusted sources". My original thinking was that untrusted sources would only be presented to parse functions. But our dialog has convinced me that we should treat the parameters to modification functions as untrusted as well. True, in some situations we will be performing unavoidable, needless re-validation. But I think it is the right tradeoff, as offering the stronger invariant has more value for users. Besides, there are workarounds for assembling a URL which trade back performance in exchange for weaker invariants. > Just to clarify what I mean: https://godbolt.org/z/qY3Y76fd9 > > urls::string_view s = "https://path?id=42&id=43"; > urls::url r = urls::parse_uri( s ).value(); > > Should the above input string not cause the invariant to be broken? (because param 'id' appears twice.) No, because the query is just a string. The interpretation of the query as "params" (an ampersand delimited list of name=value pairs) is an HTTP thing (and it has also spread to some other domains). When the query is used this way, duplicates are allowed. So the invariant is preserved in this case. It is always a valid URL. It might not be valid for custom schemes though. I could invent the "boost" scheme which requires that keys used in query parameters are unique. But there is no way for Boost.URL to enforce this (see my previous message regarding "mailto"). Users are provided with the tools to further specialize the generic URLs into custom schemes in a way that can preserve scheme-specific invariants. Thanks

5 11

Re: [boost] Boost.URL review
by Zach Laine 24 Aug '22

24 Aug '22

On Mon, Aug 22, 2022 at 9:41 AM Zach Laine <whatwasthataddress(a)gmail.com> wrote: > > On Sun, Aug 21, 2022 at 6:47 PM Vinnie Falco <vinnie.falco(a)gmail.com> wrote: > > > > On Sun, Aug 21, 2022 at 4:03 PM Zach Laine <whatwasthataddress(a)gmail.com> wrote: > > > ... > > > > It wouldn't be a proper review without at least one person forgetting > > to use "Reply All" :) So here it is, from Zach. I have quoted his > > message: > > > > > This makes sense. The larger issue I'm raising is that a user is > > > going to have a harder time remembering this, if the names don't > > > match. A way to mitigate that is to make this naming rationale > > > explicit, so that user > > > expectations will adjust. I didn't get that this was the pattern from > > > the documentation, though I think you did say it. A call-out > > > specifically noting this would porbably be enough. > > > > Yeah we can do that. > > > > > Hm. By lazy flat_map, do you mean that the positions are computed one > > > step at a time during iteration? is that why they have > > > forward_iterators? > > > > Yes. operator++ and operator-- intelligently find the next param or > > segment using a non-validating parse algorithm (since we know that the > > underlying character buffer has already been validated). > > > > > If so, please document that. It might make a > > > pretty big difference whether I decided to keep using a params object > > > that I need to update a lot, or copy everything over to a proper > > > flat_map or vector to do my work. > > > > Yes I agree, but I thought we did document that. Here, no? > > > > <https://master.url.cpp.al/url/ref/boost__urls__params_encoded_view.html#url…> > > > > We could probably add something here too: > > > > <https://master.url.cpp.al/url/ref/boost__urls__params_encoded_view/iterator…> > > Right, and I understood how those worked. However, I had no idea that > the owning containers were based on the same lazy approach (unless I > misunderstand what you said in this thread...). > > > > This works for an immutable container. It is *very surprising* for a > > > mutable one. The rest of my notes about the containers are predicated > > > on a false assumption that mutability implied non-laziness in the > > > container itself. I think all of these container would be better off > > > referred to as views, and made immutable. All you really need to > > > support is begin/end, plus maybe key-lookup. I suggest you drop the > > > mutability entirely. With mutability, these containers are trying to > > > do two things at once. > > > > Well, that's another way to do it and it certainly would remove > > complexity and simplify the names. Right now we have 8 containers: > > > > segments > > segments_view > > segments_encoded > > segments_encoded_view > > params > > params_view > > params_encoded > > params_encoded_view > > > > All of these containers reference the underlying character buffer of > > the URL (although you can also construct the views from separately > > parsed string views but that's basically the same thing in terms of > > the ownership model being reference-like). > > > > If we remove mutability as you suggest, then we can cut the number of > > containers down to 4: > > > > params_view > > segments_view > > params_encoded_view > > segments_encoded_view > > > > But this leaves us with a problem - how do you modify individual path > > segments and query params? > > Easy -- you just copy them into a flat_map, vector, or whatever. The > 90% use case is just to look at the result of the parse. Also making > the parsing library responsible for mutations you do to the data later > seems unnecessary. If you provide parsing -- including the lazy views > -- and then writing of URLs, and the writing can be done via a generic > container, that's all your library needs, IMO. To be clear: > > Parsing -> lazy_containers -> user mutations outside of Boost.URL -> > writing URLs > > That last step only needs to understand the different kinds of data it > might write, abstractly. That abstraction might be a params_view, a > flat_map, or a std::vector<std::pair<strd::string, std::string>>. > > > How do you do position-based insert and > > erase of elements using an iterator? These features seem kind of > > useful, I'm not sure I want to lose them. > > You won't lose them, because the STL exists. > > > > > Did you mean that it "doesn't" have proper error reporting? Where do > > > > you see this lack of reporting (if that's what you meant)? > > > > > > By "proper" I mean that if the parse fails it should provide an > > > end-user-friendly diagnostic, and optionally quote the line and put a > > > caret under the position that the parse failed (a la the Clang > > > diagnostics). Your non-end-user error reporting is fine. > > > > Oh... hmm.. well, each rule does leave the iterator out-param at the > > spot where an error occurred. But I'm not quite sure how to emit the > > diagnostic - surely you aren't suggesting we write to std::cout or > > std::cerr? Are we using curses or something? Or maybe just plain ASCII > > art? How exactly would this look? > > The diagnostics go to some optional output channel, which might be a > stream or file or logger. If you don't set the output, the diagnostic > is never generated. You set this output in the call to parse*(), as > an optional param -- usually a std::function<void(std::string const > &)>. > > > Thanks to Peter's work on boost::system::error_code, when a parse > > error occurs the error is augmented with file name, line, and function > > name information. And I have worked on adding visualizers so these are > > nicely presented in the debugger. So at least we have that. > > That's really nice for programmers, but it leaves the programmer with > a lot of work to do to present the error to the user. > > Zach Forgot to reply-all *again*. Zach

4 11

boost.url is accpted into boost
by Klemens Morgenstern 24 Aug '22

24 Aug '22

Boost.URL is ACCEPTED into boost without any conditions. The review recommendations were 6 ACCEPT, 1 CONDITIONAL ACCEPT and 1 REJECT. Many points were raised as part of the review, based on which I recommend doing the following: - reviewing which types/functions in the lower-level API are exposed and which ones are not (Ruben Perez' condition) - reviewing pct_encoded_view & encoded_path (as mentioned by Дмитрий) - adding more verbose docs without relying on knowledge of RFC3986 on the user side (mentioned by Richard) - adding more simple examples (recommended by Jon Kalb) - consider adding IRI support in the future (recommended by Rainer) Alan & Vinnie have been very responsive and read all the reviews, which makes it unnecessary to summarize everything. There is only one real point of controversy, namely the only reason for rejection written by Phil Endecott: > More fundamentally, I'm not convinced that this allocation-avoiding > view-rich API style is best for a general purpose vocabulary type > like URL. URL should be a "Keep It Simple" type, accessible to > C++ beginners, with straightforward lifetime semantics. Compare with > std::filesystem::path, which does not try to optimise by actually > being a view. > > Because of these concerns about the basic API design, I believe that > the library should be REJECTED. This argument has however been directly addressed in other reviews, with Seth considering it as an appropriate choice: > Like some of the earlier reviewers, I also have some concerns over introducing > non-owning vocabulary type. I do appreciate the design choice though, because I > think it ultimately fits the Boost ecosystem best - favoring > performance/versatility over ease of use. While Zach Laine considers this as the correct choice: > Overall, I quite like the design. Lazy, lazy, lazy. The fact that > only views into the underlying sequence are returned from the parse is > great, and the fact that there are both owning (url::url and > fixed-capcity owning url::static_url) and view versions of the URL > representations is exactly what I would need as a user. For my decision, the first factor is that more people approved of this particular design choice than disapproved. Secondly it seems to me that the C++ ecosystem is moving towards more view-support. The potential pitfalls can be addressed by coding guidelines (e.g. don't use auto) and do not need to be addressed by constraining libraries. Thus in finding a result, I did not consider Phil's argument to outweigh the other reviews which were considering it an acceptable choice. Other issues were the documentation, the confusion about encoded and decoded data and IRI support. Neither did lead to a rejection, so I'll leave it up to Alan's & Vinnie's discretion to review & address all the raised issues and to engage in further discussion on the mailing lists. Thank you Dmitry, Jon, Phil, Rainer, Richard, Ruben, Seth and Zach for your reviews and Alan & Vinnie for providing yet another boost library.

2 1