- Boost - lists.boost.org

exception handling anti-patterns in documentation
by Andrzej Krzemienski 30 May '22

30 May '22

Hi Everyone, I wanted to share an observation. A number of Boost libraries, in documentation, when it comes to demonstrating that a given function throws an exception, has a code sample like the following: try { > lib::function(); > } > catch(lib::exception const&) { > // handle failure > } > This applies for instance to: * Boost.Lexical_cast: https://www.boost.org/doc/libs/1_79_0/doc/html/boost_lexical_cast/examples.… * Boost.Optional: https://www.boost.org/doc/libs/1_79_0/libs/optional/doc/html/boost_optional… * Boost.URL, waiting in the review queue: https://master.url.cpp.al/url/parsing/url.html What bothers me is that this is a very dangerous antipattern related to handling exceptions. If there is any benefit to be had from exceptions, it is that they separate the "positive" control flow form the "negative" (failure handling) one. Whereas in the examples like the one above it is quite the opposite: we are trying to catch the exception as soon as it is thrown. This is usually the cause of bugs and resource leaks. Such examples also lead many young programmers into thinking that exceptions are difficult or illogical. But they are not illogical: it is antipatterns like the one above that, which are working against the design of the exception handling mechanism, that confuse the programmers and enforce bad habits. One could say that the goal of the documentation is to demonstrate, in an efficient way, the interface of the library: not to teach how to program. But I would disagree. I have always considered Boost libraries to be more than that. I personally learned a lot about how to program, organize my code and organize my thinking about abstractions from reading the documentation of Boost libraries. Boost is a place to promote better programming practices. And these "catch immediately after throw and swallow" examples just do a disservice to the community. I would like to know if there are more people in this list that feel the same way. Maybe we could come up with a systematic way of demonstrating a throwing library interface, or guidelines for documentation authors. Regards, &rzej;

8 18

nullptr vs 0
by Peter Dimov 30 May '22

30 May '22

Compilers apparently are warning on the use of 0 as a null pointer constant, suggesting we use nullptr instead. But compilers don't know that we support C++03 where nullptr isn't a thing. Case in point: https://github.com/boostorg/throw_exception/pull/22 ifdef-ing every use of nullptr is unwieldy, so maybe we need BOOST_NULLPTR added to Boost.Config? That would expand to nullptr when it's supported, and 0 otherwise.

15 38

Re: [boost] Boost MySQL Review
by Ruben Perez 28 May '22

28 May '22

CC'ing the ML, as you just replied to me. On Sun, 22 May 2022 at 14:51, Seth <bugs(a)sehe.nl> wrote: > > Rubén, quick responses to your questions: > > >> TARGET_LINK_LIBRARIES(reviewtest Boost::mysql) > > > > This shouldn't be the case at all. The second line should be sufficient. > > Could you please open an issue against the repo and post your > > CMakeLists.txt, please? > > I think I understand what I did wrong/differently. I've filed the description as https://github.com/anarthal/mysql/issues/97 for potential improvement. I think there was an error in your link_libraries command, please have a look at the issue reply. I will leave the issue open to provide an example, though. > > > Implementing MySQL semantics implies the following: > > > > value("1") == value(1) // true > > value(1.0) == value(1) // true > > > > Which reminds me of the kind of patterns that have brought us > > pain in PHP and JS, and would prefer to avoid. Do you think > > "placing values in containers" is something that adds value? > > Obviously Mysql's type system is not something we want to embrace. In fact, my stance is that `mysql::value` should be for marshaling, not as a vocabulary type. As such, I do *not* think that general purpose support for "placing values in containers" adds value. > > Implementation-specific specialized containers like `mysql::row` can already handle the type appropriately. In fact I note that it does more harm than good because of the... surprising (some would say, broken) semantics. Again, for library internal use those are fine, but end-users should not get a false sense of "normalcy". > > Perhaps the library could document some utilities like `value_equal` or `value_hash` for explicit use. I'm reluctant to remove operator== and operator!=. I'd say their behavior matches the one for variants and can be useful. I see your point and I'm completely up to removing the other relational operators. > > >> > >> 3. The documentation claim "Values do not define `std::hash` yet, as > >> `boost::string_view` does not support it" is out of date. `hash_value` has > >> been added in 1.69.0 > > > > I can see a definition of hash_value, and from what I've read this makes > > it compatible with Boost containers, but the std::hash specializations are > > in a #ifdef 0 block. AFAIR std::hash doesn't use hash_value, and > > std::unordered_set<boost::string_view> fails to compile. The same > > applies for std::unordered_set<boost::mysql::value>. > > > Boost support is hash support, and it's trivial to `template <> struct std::hash<mysql::value> : boost::hash<mysql::value>` if desired. Missed that one. Thanks. Taking your comments into account, however, I'll likely won't provide it, though. > > >> 4. On that note, consider `boost::core::string_view` for better interop > >> ("lingua franca" string_view). If you're interested I just did an exhaustive > >> side-by-side comparison of features and observable behaviour` which I can > >> share. > > > > I wasn't aware of this class. Doing a quick grep in Beast, I've found it does > > use it, but it seems located in a detail/ folder > > (<boost/core/detail/string_view.hpp>). > > Could you please provide me with your comparison and any extra info I should > > be aware of? (BTW std::unordered_set<boost::core::string_view> doesn't compile > > either). > > My impact analysis consists of a suite of Very Ugly(TM) test cases that have lead to differences marked "OBSERVABLE" in the test code. It's not intended for publication, and I was lazy by using C++17 features. However, it has the "code talks" quality in that it leaves little to the imagination. So, please remember to wear goggles: https://github.com/sehe/beast/blob/core_string_view/test/beast/core/string.… > > The main area I didn't compare was constexpr-ness. Anecdotally I got the impression that core::string_view is more consistent in constexpr-ification. In fact my analysis uncovered a bug in Boost Utility https://github.com/boostorg/utility/issues/94 > > Some other differences have been addressed for an upcoming Boost release. The related tickets can be found in this cppslack thread: https://cpplang.slack.com/archives/CD7BDP8AX/p1651467929687909 I don't think any of these differences impact this library. I've raised https://github.com/anarthal/mysql/issues/98 to address it. It still freaks me out a little including a header labeled as "detail" - why is it not available as <boost/core/string_view.hpp> rather than <boost/core/detail/string_view.hpp>? It looks like if the class wasn't supposed to be used in public interfaces. > > > I'd say that the pipeline interface > > (https://github.com/anarthal/mysql/issues/76) > > may be a better interface than multi-statement for this use case. > > I think what I tried to describe is what you call the pipeline interface. So +1 > > > Note that prepared_statement's get allocated a server-side ID, which > > we expose via the prepared_statement::id() function. Something that > > could be useful for that use case is closing a statement without a > > prepared_statement object, just using its ID. > > > It may be worth adding a hint about the native_handle with the current warning about prepared_statement's destructor. > > > > > Something I noticed while implementing these erased types is that > > having prepared_statement and resultset be I/O objects makes > > the erased interface much harder. As I've mentioned in other > > review responses, I'm thinking on replacing the following signatures: > > > > prepared_statement::execute(const statement_params<ValueIterator>&) > > prepared_statement::close() > > resultset::read_one(row&) > > > > By functions in the connection: > > > > connection::execute_statement(const prepared_statement&) > > connection::close_statement(const prepared_statement&) > > connection::read_row(resultset&, row&) > > > > Where prepared_statement and resultset would be non-templated > > types. What do you think of this API? > > I can see how it simplifies things. It feel a bit C-api to me. But I have no strong opinion either way. > It's what you think is best for users in the end. Regards, Ruben.

3 2

Re: [boost] need help with building QuickBook docs
by Andrzej Krzemienski 25 May '22

25 May '22

I managed to figure out that it is this command that triggers the error: "xsltproc" -v --stringparam boost.defaults "Boost" --stringparam > boost.libraries "../../../../libs/libraries.htm" --stringparam boost.root > "../../../.." --stringparam chapter.autolabel "0" --stringparam > chunk.section.depth "8" --stringparam generate.section.toc.level "1" > --stringparam toc.max.depth "2" --stringparam toc.section.depth "2" --path > "../../bin.v2" --path "../../bin.v2/libs/optional/doc" --xinclude -o > "../../bin.v2/libs/optional/doc/optional.docbook" > "/home/andrzej/Repos/boost/tools/boostbook/xsl/docbook.xsl" > "../../bin.v2/libs/optional/doc/optional.xml" > I am using xsl files in boost/tools/boostbook/xsl from Boost repo. The contents of optional.xml are attached. It looks like QuickBook went fine. Well, this is now manageable. I got it to the point where QuickBook can detect bugs in my *.qbk files, and I can commit changes in documentation without the fear of introducing broken sources, but without the comfort of seeing the end results. I can then watch them in https://www.boost.org/doc/libs/develop/ Regards, &rzej; pon., 23 maj 2022 o 22:54 Andrzej Krzemienski <akrzemi1(a)gmail.com> napisał(a): > > > pon., 23 maj 2022 o 22:11 Andrey Semashev via Boost <boost(a)lists.boost.org> > napisał(a): > >> On 5/23/22 22:14, Andrzej Krzemienski wrote: >> > >> > >> > pon., 23 maj 2022 o 00:30 Andrey Semashev via Boost >> > > >> > > Are you using the latest built QuickBook >> > > executable? >> > >> > How can I check my version of QuickBook? >> >> quickbook --version >> >> But if you need to ask, it is probably better to just build it from >> sources. >> > > $ quickbook --version > Quickbook Version 1.7.2 (Boost 1.76) > > > >> >> > Also, make sure you don't have some stale content in bin.v2. Your >> log >> > doesn't indicate that QuickBook was run, probably because the >> BoostBook >> > document was already there. >> > >> > I now removed everything from folder bin.v2 before running `b2 doc`. I >> > still get the same result. >> > Maybe there is a way to see intermediate results, such as the generated >> > BoostBook format? >> >> The intermediate results can be found in bin.v2/libs/optional. >> > > So, the only intermediate result I found there is > bin.v2/libs/optional/doc/optional.xml (enclosed). > It has: > > <!DOCTYPE library PUBLIC "-//Boost//DTD BoostBook XML V1.0//EN" " > http://www.boost.org/tools/boostbook/dtd/boostbook.dtd"> > > Which is the output of quickbook, I suppose. > > I am using libxslt-1.1.35-1.fc36.x86_64, and it seems the lowest one that > I can get on my Fedora. > > # xsltproc --version > Using libxml 20914, libxslt 10135 and libexslt 820 > xsltproc was compiled against libxml 20912, libxslt 10135 and libexslt 820 > libxslt 10135 was compiled against libxml 20912 > libexslt 820 was compiled against libxml 20912 > > Regards, > &rzej; > > >> _______________________________________________ >> Unsubscribe & other changes: >> http://lists.boost.org/mailman/listinfo.cgi/boost >> >

2 2

Need help with SOCKS-Asio library
by Alan de Freitas 25 May '22

25 May '22

Hi! First of all, I'd like to apologize for asking for help last time during the Boost.MySql review. Terrible timing. :) So considering our recent discussions about the future of Boost, where providing more protocol implementations comes up frequently, we started this SOCKS <https://en.wikipedia.org/wiki/SOCKS> library I would like to "re-announce." The library is based on Asio and already contains the most relevant operations. There's also a client example and minimal proxy server example you can already use with your browser. At this point, I'd like to know if any kind soul, including the Asio experts, could have a quick look (as quick as you want) and provide some feedback regarding the implementation or even just the suitability of the library direction. Repo: https://github.com/alandefreitas/socks_proto Docs: https://develop.socks-proto.cpp.al/ #boost-socks on Slack Thanks! -- Alan Freitas https://github.com/alandefreitas

5 7

What do I need Boost for?
by Andrzej Krzemienski 25 May '22

25 May '22

Hi Everyone, In connection with the recent discussions on the future of Boost, I would like to offer a perspective of someone who is a consumer of open-source (but also commercial) libraries. In my work I often find myself in a situation where I need to use a new library for: * logging * unit testing * communication with DB * parsing requests XML, or simplified XML (without Unicode or less popular XML features), JSON * validating requests against XML/JSON schemas * HTTP communication * etc... I do not necessarily need Boost libraries for that. I need libraries that satisfy my criteria for correctness, safety, performance, and conformance with a certain programming style (e.g., a library cannot force me to inherit all my classes from class `Object`). These libraries do not necessarily have to come from one source. But they have to *interoperate* with one another. When I select a new library my concerns are usually the same: 1. Does the author understand the problem of exception safety and failure safety in general? The declaration "this library is exception safe" is an indication that the author at least heard about the term, but is still insufficient. I become convinced when I see a listing: these functions offer commit-or-rollback guarantee, these functions offer basic failure-safety guarantee, these functions offer other guarantees, like never-empty for some implementations of variants. 2. Does the author understand the problem of thread-safety? Similarly here, the note "this library is thread-safe" at least indicates that the author is aware of thread-safety problems. I feel more confident when I see the listing: "these functions can be called simultaneously from multiple threads", or similar declarations. 3. Does the library use the established patterns in C++, like value semantics, strong types? 4. Will the library give me the control over resource management? Or will it allocate like 30 threads behind my back, when I least expect it? 5. Will it interoperate with other libraries that I choose for other tasks? 6. Is the library actively maintained? 7. Is the library *decently* documented? (Not only the repetition of declarations from the header files.) 8. Did the author study the domain deeply, and has come up with the optimum concepts and the interface? Boost is often my first place to go. Because I have a high level of confidence that my expectations will be satisfied. But I do not always prefer Boost. For instance for unit tests I go to Google.Test, for instance, because I strongly disagree with the choice for how the precision for floating-point comparisons is silently chosen in a non uniform way. My confidence in boost stems from the review process and the assembly of experts that I trust. One could say that Boost offers a certification mechanism, even if informal. I do not care that much if all libraries come from one source, but I would love there to be a common certification body (even if informal) that would say "this library meets a certain set of criteria". I desperately needed Boost in the past because it offered a really basic fundamental primitives that were missing from the Standard library, as well from the language: optional, variant, tuple, any, shared_ptr, but also BOOST_FOREACH, BOOST_AUTO. You could call the types from the list the "vocabulary types": they can be used in other third party libraries, and therewith different libraries can interoperate seamlessly owing to these types. Today, these types from Boost do not necessarily fill this role anymore because we have the std equivalents. But Boost.ASIO can be thought of as such a building block: it can be used as a base for many higher-level libraries that interoperate owing to the same asynchrony model. If I, a library consumer, were to answer the question what I would like Boost to be, that would be: 1. The certification role: to educate what a high standard library means, to offer help and tools for library developers to achieve this level (not necessarily as a Boost library) and to issue opinions on the libraries. Boost to a great extent already implements this. 2. To focus on foundational libraries: those that can help other libraries interoperate. Like the one big decimal library. An interesting counter example to using vocabulary types is Boost.JSON implementing its own string type for efficiency. I was already convinced that it was the right decision to implement a custom string. But it illustrates something important: we cannot have the one string class. I guess where there is a need to manage resources (like string manages memory internally) there cannot be the one vocabulary type. Regards, &rzej;

2 1

need help with building QuickBook docs
by Andrzej Krzemienski 23 May '22

23 May '22

Hi Everyone, This is to ask for help with building QuickBook documentation, but also to spill my frustration. I would like to offer some feedback in connection with the recent "the future of Boost" threads: about the entry barrier for someone who wants to contribute to Boost. I maintain Boost.Optional. Getting my environment to the point where I was actually able to apply modifications to the official repo took some time, but it has been working fine since. Modifying the source files, so that they compile on multiple compilers, platforms and versions, including C++03 is the easiest part. Testing is a bit harder, but basically works also: I go to folder `test`, run the `b2` command, and it is done. But updating the documentation is the most difficult task for me. Not the content -- this is easy -- but the process/pipeline/framework. I inherited the documentation in QuickBook format, which as I understand was the recommended way for providing documentation for Boost libraries. Maybe it still is? (but I see more and more docs being rewritten to AsciDocs.) I understand that library authors have the freedom to choose their own format of docs for their libraries, but does boost recommend (as opposed to enforcing) one format? Anyway, in order to test if my QuickBook dos are correct, I need to build HTML version from the sources. Sometimes it "just works" with `b2` command, but when something goes wrong, this becomes a pain. I cannot just call `b2` for my library, because apparently my Fedora comes with preinstalled version of b2 (4.4) which appears to be incompatible with something in Boost libraries sources, as I get error: error: mismatched versions of B2 engine and core > error: B2 engine (b2) is 4.4.1 > error: B2 core (at /home/andrzej/Repos/boost/tools/build/src) is 4.9-git > Ok, I learned to live with it and to launch b2 from tools/build. It used to work for a while, but yesterday when I did a `git pull --rebase` on Boost repo, it no longer works. I get an error: Jamfile.v2:45: Unescaped special character in argument > <format>pdf:<xsl:param>boost.url.prefix= > http://www.boost.org/doc/libs/release/libs/optional/doc/html > Performing configuration checks > > - default address-model : 64-bit (cached) [1] > error: at /home/andrzej/Repos/boost/tools/build/src/kernel/modules.jam:106 > error: Unable to find file or target named > error: '/boost/architecture//mips' > error: referred to from project at > error: '.' > Ok, I can remove line `: /boost/architecture//mips` from file `boostcpp.jam`. (Am I supposed to do it before using `b2`?). But then, when trying to build the docs, I get a lot of errors, saying: runtime error: file > /home/andrzej/Repos/boost/tools/boostbook/xsl/annotation.xsl line 432 > element element > xsl:element: The effective name '' is not a valid QName. > (Full error message is enclosed.) I have no clue which element in the pipeline is causing or reporting this. QuickBook, BoostBook, DocBook, xsltproc? The process involves many components, at least according to this diagram: https://svn.boost.org/trac10/wiki/BoostDocs/GettingStarted Is this something obvious that I am missing? Is this a good idea, or a recommended practice to use QuickBook nowadays for Boost docs? I feel comfortable fixing or writing C++ code, and I am willing to devote some of my limited time to do this. But fighting with QuickBook documentation is something I would not rather do. I have a small library that I considered submitting to Boost: https://github.com/akrzemi1/markable But as long as it is on my GitHub page, it is simple to maintain, test and document. If I imagine how many nights I will have to spend making the documentation compatible with Boost, or getting all the b2-related stuff to work, I feel discouraged. Maybe I am being unfair to `b2` now. But it is certainly my personal emotional impression. I do not mind the current Boost website format or the underlying technology. I do not mind the mailing list format. My primary obstacle is this perceived or real difficulty to get a boost library into a shape where it just builds, including its documentation. My ideal as a Boost library maintainer, is that I just provide the source for the library and the documentation, and anything else "just works". Maybe this is not achievable, but the closer to this ideal, the more encouraged I would be to consider authoring or maintaining a library. Any help in fixing my build problem is most welcome. Regards, &rzej;

7 16

MySql review
by Phil Endecott 23 May '22

23 May '22

Here is my review of Ruben Perez's proposed MySql library. Background ---------- I have previously implemented C++ wrappers for PostgreSQL and SQLite, so I have some experience of what an SQL API can look like. I know little about ASIO. I have also recently used the AWS SDKs for C++ and Javascript to talk to DynamoDB; this has async functionality, which is interesting to compare. I confess some minor disappointment that MySql, rather than PostgreSQL or SQLite, is the subject of this first Boost database library review, since those others have liberal licences that are closer to Boost's own licence than MySql (and MariaDB). But I don't think that should be a factor in the review. Trying the library ------------------ I have tried using the library with - g++ 10.2.1, Arm64, Debian Linux - ASIO from Boost 1.74 (Debian packages) - Amazon Aurora MySql-compatible edition I've written a handful of simple test programs. Everything works as expected. Compilation times are a bit slow but not terrible. The remainder of this review approximately follows the structure of the library documentation. Introduction ------------ I note that "Ease of use" is claimed as the first design goal, which is good. I feel that some mention should be made of the existing C / C++ APIs and their deficiencies. You should also indicate whether or not the network protocol you are using to communicate with the server is a "public" interface with some sort of stability guarantee. (I guess maybe it is, if it is common to MySql and MariaDB.) Tutorial -------- The code fragments should start with the necessary #includes, OR you should prominently link to the complete tutorial source code at the start. You say that "this tutorial assumes you have a basic familiarity with Boost.Asio". I think that's unfortunate. It should be possible for someone to use much of the library's functionality knowing almost nothing about ASIO. Remember your design goal of ease-of-use. In fact, it IS possible to follow the tutorial with almost no knowledge of ASIO because I have just done so. You have this boilerplate at the start of the tutorial: boost::asio::io_context ctx; boost::asio::ssl::context ssl_ctx (boost::asio::ssl::context::tls_client); boost::mysql::tcp_ssl_connection conn (ctx.get_executor(), ssl_ctx); boost::asio::ip::tcp::resolver resolver (ctx.get_executor()); auto endpoints = resolver.resolve(argv[3], boost::mysql::default_port_string); boost::mysql::connection_params params ( argv[1], // username argv[2] // password ); conn.connect(*endpoints.begin(), params); // I guess that should really be doing something more // intelligent than just trying the first endpoint, right? I would like to see a convenience function that hides all of that: auto conn = boost::mysql::make_connection( ...params... ); I guess this will need to manage a global, private, ctx object or something. There are various possibilities for the connection parameters to pass to that, e.g. connection_params params = { .hostname = ".....", .port = 3306, // why is that a string in yours? .user = "admin", .password = "12345", .dbname = "foo" }; make_connection(params); Or make_connection("mysql://admin:12345@hostname:3306/dbname"); Now... why the heck does your connection_params struct use string_views? That ought to be a Regular Type, with Value Semantics, using std::strings. Is this the cult of not using strings because "avoid copying above all else"? Another point about the connection parameters: you should provide a way to supply credentials without embedding them in the source code. You should aim to make the secure option the default and the simplest to use. I suggest that you support the ~/.my.cnf and /etc/my.cnf files and read passwords etc. from there, by default. You might also support getting credentials from environment variables or by parsing the command line. You could even have it prompt for a password. Does MySQL support authentication using SSL client certs? I try to use this for PostgreSQL when I can. If it does, you should try to support that too. About two thirds of the way through the tutorial, it goes from "Hello World" to retrieving "employees". Please finish the hello world example with code that gets the "Hello World" string from the results and prints it. Queries ------- I encourage you to present prepared queries first in the documentation and to use them almost exclusively in the tutorial and examples. You say that "client side query composition is not available". What do you mean by "query composition"? I think you mean concatenating strings together'); drop table users; -- to form queries, right? Is that standard MySql terminology? I suggest that you replace the term with something like "dangerous string concatenation". In any case, that functionality *is* available, isn't it! It's trivial to concatenate strings and pass them to your text query functions. You're not doing anything to block that. So what you're really saying is that you have not provided any features to help users do this *safely*. I think that's a serious omission. It would not be difficult for you to provide an escape_for_mysql_quoted_string() function, rather than having every user roll their own slightly broken version. IIRC, in PostgreSQL you can only use prepared statements for SELECT, UPDATE, INSERT and DELETE statements; if you want to do something like ALTER TABLE a ALTER COLUMN c SET DEFAULT = ? or CREATE VIEW v as SELECT * FROM t WHERE c = ? then you are forced to do string concatenation with escaping of parameters. But according to https://dev.mysql.com/doc/refman/8.0/en/sql-prepared-statements.html, MySQL allows more statements to be prepared. In what cases can prepared statements not be used? Perhaps non-prepared statements really can be hidden as mysql::unsafe::statement? Regarding your prepared statements: tcp_ssl_prepared_statement is verbose. Why does the prepared statement type depend on the underlying connection type? I have to change it if I change the connection type?! If that's unavoidable, I suggest putting a type alias in the connection type: connection_t conn = ....; connection_t::prepared_statement stmt(.....); Does MySql allow numbered or named parameters? SQLite supports ?1 and :name; I think PostgreSQL uses $n. Queries with lots of parameters are error-prone if you just have ?. If MySql does support this, it would be good to see it used in some of the examples. Invoking the prepared statement seems unnecessarily verbose. Why can't I just write auto result = my_query("hello", "world", 42); ? That's a variadic operator(). In my PostgreSQL library, I chose to specify the query parameter types as template parameters to the query (prepared statement) type: Query<std::string, std::string, int> my_query("....."); Query's operator() takes exactly those types (maybe with some TMP to use std::string_view where std::string is specified). With hindsight, a default where all parameters are strings would have been useful in practice. I also added Query variants where the result is expected to be - A single value, e.g. a SELECT COUNT(*) statement. - Empty, e.g. INSERT or DELETE. - A single row. - Zero or one rows. - A single column. This is just syntactic sugar but it makes life easier. For example: SingletonQuery<int, string> sessions_for_user("SELECT COUNT(*) FROM sessions WHERE username = ?"); // int is the return type, string is the one parameter type. auto n_sessions = sessions_for_user("phil"); I don't see anything about the result of an INSERT or UPDATE. PostgreSQL tells me the number of rows affected, which I have found useful to return to the application. resultset, row and value ------------------------ I'm not enthusiastic about the behaviour nor the names of these types: - resultset is not a set of results. It's more like a sequence of rows. But more importantly, it's lazy; it's something like an input stream, or an input range. So why not actually make it an input range, i.e. make it "a model of the input_range concept". Then we could write things like: auto result = ...execute query... for (auto&& row: result) { ... } - row: not bad, it does actually represent a row; it's a shame it's not a regular type though. - value: it's not a value! It doesn't have value semantics! I guess it is time to address the string_view issue..... I just modified one of my example programs to copy from the string_view in the variant into a std::string, and it slowed the program down by... no, it actually went fractionally faster. The use of string_views in the value variant seems to me to be a premature optimisation. There are significant disadvantages to using string_view here. Your default API should have value semantics. If you want, provide another "low copying" API for users who want it with names that make it clear what they are doing. The problem is that the lifetime of the string_view is tied to the row in a non-obvious way. I'm also uncertain that a variant for the individual values is the right solution here. All the values in a column should have the same type, right? (Though some can be null.) So I would make row a tuple. Rather than querying individual values for their type, have users query the column. MySQL to C++ mapping reference ------------------------------ It seems odd that MySQL small integers all map to C++ 64-bit types. I use NUMERIC quite a lot in PostgreSQL; I don't know if the MySql type is similar. I would find it inconvenient that it is treated as a string. Can it not be converted to a numeric type, if that is what the user wants? Need to consume the entire resultset ------------------------------------ I seem to get an assertion if I fail to read the resultset (resultset.hpp:70). Could this throw instead? Or, should the library read and discard the unread results in this case? Going Async ----------- I have tried this with both callbacks and C++20 coroutines and it seems to work as expected. I'm actually delighted to see the support for coroutines here. But the lack of protocol support for multiple in-flight queries immediately becomes apparent. It almost makes me question the value of the library - what's the point of the async support, if we then have to serialise the queries? Should the library provide this serialisation for us? I.e. if I async_execute a query while another is in progress, the library could wait for the first to complete before starting the second. Or, should the library provide a connection pool? (Does some other part of ASIO provide connection pool functionality that can be used here?) Transactions ------------ I have found it useful to have a Transaction class: { Transaction t(conn); // Issues "BEGIN" .... run queries .... t.commit(); // Issues "COMMIT" } // t's dtor issues "ROLLBACK" if we have not committed. Nested transactions are an interesting problem. Trademark --------- Klemens Morgenstern makes the point that MySql is a trademark of Oracle. Calling this "Boost.MySql" doesn't look great to me. How can you write "The Boost MySql-compatible Database Library" more concisely? Conclusion ========== Good points: It actually works! Great to see coroutines! Bad points: The resultset / row / value type system is a poor design. A lot of boilerplate could be hidden from many users. Connection pool may be required for useful performance. Several features should be re-worked to be secure by default. Use of MySQL trademark needs to be considered. Overall, I think this proposal needs a fair amount of API re-design and additional features to be accepted, and should be rejected at this time. It does seem to be a good start though! Thanks to Ruben for the submission. (I have spend one working day on this review. I have not looked at any of the code.) Regards, Phil.

13 18

Re: [boost] need help with building QuickBook docs
by Andrzej Krzemienski 22 May '22

22 May '22

niedz., 22 maj 2022 o 17:33 <pbristow(a)hetp.u-net.com> napisał(a): > I don’t see where annotation.xsl comes from, or what is should contain > (but it looks as though it contains something wrong!!). > > > > Can you share it with us? And any metadata? > The file comes from the boostbook repo: https://github.com/boostorg/boostbook/blob/develop/xsl/annotation.xsl I applied no modifications. > > > None of this looks familiar to me from generating Quickbook docs ☹ > > > > Paul > > > > PS I also view items reported as cached with suspicion, as a result of > being bitten before. > > > > Especially when there are two difference versions. > > > > > https://stackoverflow.com/questions/5709105/how-do-i-force-bjam-in-boost-to… > may help. > Thanks for the idea. I'll try it later, in the evening. Regards, &rzej; > > > > > *From:* Andrzej Krzemienski <akrzemi1(a)gmail.com> > *Sent:* 22 May 2022 15:45 > *To:* Boost mailing list <boost(a)lists.boost.org> > *Cc:* pbristow(a)hetp.u-net.com > *Subject:* Re: [boost] need help with building QuickBook docs > > > > > > > > niedz., 22 maj 2022 o 15:52 Paul A Bristow via Boost < > boost(a)lists.boost.org> napisał(a): > > > -----Original Message----- > > From: Boost <boost-bounces(a)lists.boost.org> On Behalf Of Andrzej > Krzemienski via Boost > > Sent: 22 May 2022 14:14 > > To: René Ferdinand Rivera Morell <grafikrobot(a)gmail.com> > > Cc: Andrzej Krzemienski <akrzemi1(a)gmail.com>; Boost Developers List < > boost(a)lists.boost.org> > > Subject: Re: [boost] need help with building QuickBook docs > > > > Thanks. That would be the most likely cause of my problems. Could you > point me to instructions for > > rebuilding the b2 engine executable? > > > > https://www.boost.org/doc/libs/1_79_0/more/getting_started/windows.html#get… > > > Section 5.1 Simplified Build From Source > > If you wish to build from source with Visual C++, you can use a simple > build procedure described in this section. Open the command prompt and > change your current directory to the Boost root directory. Then, type the > following commands: > > bootstrap > .\b2 > The first command prepares the Boost.Build system for use. The second > command invokes Boost.Build to build the separately-compiled Boost > libraries. Please consult the Boost.Build documentation for a list of > allowed options. > > https://www.boost.org/doc/libs/1_79_0/tools/build/doc/html/index.html > > > https://www.boost.org/doc/libs/1_79_0/tools/build/doc/html/index.html#bbv2.… > > Run either .\bootstrap.bat (on Windows), or ./bootstrap.sh (on other > operating systems) > > (which may be too simple. By default, Windows 10 and up, probably don't > provide cmd.exe, preferring the fancy new command shell Windows > PowerShell) > > b2 -v will show the current version. > > You will probably want to ensure that b2.exe is visible globally, perhaps > by adding to PATH. Perhaps using the Windows Environment Variables editor? > > HTH > > Paul > > > > Thank you. This seems to remove the first obstacle on my way to building > Boost.Optional docs. > > I am now using B2 Version 4.9. OS=LINUX. I get the error from another > tool, I suppose: > > > > [2] > gcc-12/release/cxxstd-0x-iso/link-static/python-3.10/threadapi-pthread/threading-multi/visibility-hidden > ...patience... > ...patience... > ...found 4351 targets... > ...updating 2 targets... > xslt-xsltproc ../../../bin.v2/libs/optional/doc/optional.docbook > runtime error: file > /home/andrzej/Repos/boost/tools/boostbook/xsl/annotation.xsl line 432 > element element > xsl:element: The effective name '' is not a valid QName. > > > > Is it DocBook, QuickBok, BoostBook? (I enclose the full error report.) I > am just running b2 on boost/libs/optional/doc from the official repo. Can > someone confirm if they see the same problem? or is it just me? > > > > Regards, > > &rzej; > > > > > > > > > > > _______________________________________________ > Unsubscribe & other changes: > http://lists.boost.org/mailman/listinfo.cgi/boost > >

1 0

[Boost MySQL] - Review Period has Ended
by Richard Hodges 21 May '22

21 May '22

Looking at the number of messages in my inbox I see that we have had a great response to the request for reviewers while I have been in transit. Thank you everyone for taking the time to participate. Thank you also to Rubén, as it is apparent from the number of github commits I have seen that he has been very responsive to review comments. Now that I am back home it will take me a little time to collate the results and present. I expect to have this done by the end of the weekend. -- Richard Hodges hodges.r(a)gmail.com office: +44 2032 898 513 home: +376 861 195 mobile: +376 380 212

1 0