June 2014 - Boost - lists.preview.boost.org

How to make bjam use gcc instead of g++ using custom compiling and linking options
by Edson Eustachio 22 Jun '14

22 Jun '14

Due to some environment settings I cannot change, I have to build a lib using bjam instead of make. In my own development environment, I have a Makefile that can successfuly build the lib with the following options: gcc -x c -fPIC -shared -w -I include -o libfmtkit.so -lm src/fm5x.c src/fmTXT.c src/fmAiq.c src/fmArray.c src/fmCalc.c src/fmCandle.c src/fmCT.c src/fmCSI.c src/fmDate.c src/fmDJ.c src/fmFile.c src/fmFin.c src/fmFS.c src/fmPNF.c src/fmStr.c src/fmTc.c src/fmTrade.c src/fmTT.c So, I want to reproduce the options above in a Jamfile The first step is replace g++ by gcc. I've already found this can be done adding the rule using gcc But when I do that I got : $ bjam error: at /usr/share/boost-build/build/project.jam:825 error: duplicate initialization of gcc with the following parameters: error: version = 4.4.3 So, I conclude I am already using gcc instead g++. Is that right? The second step is overwrite the options -fPIC -shared -w to the my Jamfile Below is the Jamfile: lib fmtkit : src/fm5x.c src/fmAiq.c src/fmArray.c src/fmCSI.c src/fmCT.c src/fmCalc.c src/fmCandle.c src/fmChartW.c src/fmDJ.c src/fmDate.c src/fmFS.c src/fmFile.c src/fmFin.c src/fmGraph.c src/fmPNF.c src/fmStr.c src/fmTT.c src/fmTXT.c src/fmTc.c src/fmTrade.c : <include>include : <link>shared; I've passed thru the boost documentation but I've not found or I'm not able to understand how to do that into my Jamfile. Does anyone have some clue of where should I find instructions to setting that? This is the output I get when I try to build the lib using the Jamfile described above: $ bjam...found 81 targets......updating 3 targets... gcc.compile.c bin/gcc-4.4.3/debug/fmChartW.oIn file included from src/fmChartW.c:7: include/fmChartW.h:163: error: expected ')' before 'hdc' include/fmChartW.h:166: error: expected ')' before 'hdc' include/fmChartW.h:175: error: expected ')' before 'hdc' include/fmChartW.h:176: error: expected ')' before 'hdc' src/fmChartW.c:13: error: expected ')' before 'hdc' src/fmChartW.c:300: error: expected ')' before 'hdc' src/fmChartW.c: In function 'fmChartScreenInit': src/fmChartW.c:323: warning: implicit declaration of function 'malloc' src/fmChartW.c:323: warning: incompatible implicit declaration of built-in function 'malloc' src/fmChartW.c:329: warning: implicit declaration of function 'free' src/fmChartW.c:329: warning: incompatible implicit declaration of built-in function 'free' src/fmChartW.c: In function 'fmChartScreenClose': src/fmChartW.c:350: warning: incompatible implicit declaration of built-in function 'free' src/fmChartW.c: At top level: src/fmChartW.c:396: error: expected ')' before 'hdc' src/fmChartW.c:604: error: expected ')' before 'hdc' "g++" -x c -O0 -fno-inline -Wall -g -fPIC -I"include" -c -o "bin/gcc-4.4.3/debug/fmChartW.o" "src/fmChartW.c" ...failed gcc.compile.c bin/gcc-4.4.3/debug/fmChartW.o... gcc.compile.c bin/gcc-4.4.3/debug/fmGraph.o ... src/fmGraph.c:2544: error: 'fmSCALETYPESEMILOG' undeclared (first use in this function) src/fmGraph.c:2546: error: dereferencing pointer to incomplete type src/fmGraph.c:2546: error: dereferencing pointer to incomplete type src/fmGraph.c:2546: error: dereferencing pointer to incomplete type src/fmGraph.c:2546: error: dereferencing pointer to incomplete type src/fmGraph.c:2546: error: dereferencing pointer to incomplete type src/fmGraph.c:2546: error: dereferencing pointer to incomplete type src/fmGraph.c:2546: error: dereferencing pointer to incomplete type src/fmGraph.c:2546: error: dereferencing pointer to incomplete type src/fmGraph.c:2547: error: dereferencing pointer to incomplete type src/fmGraph.c:2551: error: dereferencing pointer to incomplete type src/fmGraph.c:2551: error: dereferencing pointer to incomplete type src/fmGraph.c:2551: error: dereferencing pointer to incomplete type src/fmGraph.c:2551: error: dereferencing pointer to incomplete type src/fmGraph.c:2551: error: dereferencing pointer to incomplete type src/fmGraph.c:2551: error: dereferencing pointer to incomplete type src/fmGraph.c:2552: error: dereferencing pointer to incomplete type src/fmGraph.c: In function 'fmGraphYpos': src/fmGraph.c:2565: error: 'fmSCALETYPESEMILOG' undeclared (first use in this function) src/fmGraph.c:2573: error: 'fmSCALETYPEINVERTED' undeclared (first use in this function) "g++" -x c -O0 -fno-inline -Wall -g -fPIC -I"include" -c -o "bin/gcc-4.4.3/debug/fmGraph.o" "src/fmGraph.c" ...failed gcc.compile.c bin/gcc-4.4.3/debug/fmGraph.o......skipped <pbin/gcc-4.4.3/debug>libfmtkit.so for lack of <pbin/gcc- 4.4.3/debug>fmChartW.o......failed updating 2 targets......skipped 1 target... Thanks in advance

1 0

big problem with dependency changes
by Robert Ramey 21 Jun '14

21 Jun '14

Currently the master branch has a few problems. Here is one typical example. My project uses the current master to provide boost libraries which I'm not working on. One file has the statement #include <boost/utility/enable_if.hpp> which fails to compile since boost/utility/enable_if.hpp has been moved to core. This raises a few questions. a) Isn't the master branch always supposed to be working? b) There are a lot of intermediate checkins which break the master. Should these be done an a branch and, once they are believed to be all correct, merged in as a group. This would support the concept a) above. c) will I have to change code in the library I maintain to manually change all occurrences of #include <boost/utility/enable_if.hpp> to #include <boost/core/enable_if.hpp> ? d) Why are the tests on the master branch passing? Robert Ramey -- View this message in context: http://boost.2283326.n4.nabble.com/big-problem-with-dependency-changes-tp46… Sent from the Boost - Dev mailing list archive at Nabble.com.

11 27

Efficient tuple implementation
by Louis Dionne 20 Jun '14

20 Jun '14

Hi, I recently discovered (or maybe not) a neat trick to implement a tuple-like container. The technique has a couple of drawbacks which I will explain later. For a real example, you can see my list implementation in Boost.Hana at [1]. Here's the idea: auto list = [](auto ...xs) { return [=](auto access) { return access(xs...); }; }; auto head = [](auto xs) { return xs([](auto first, auto ...rest) { return first; }); }; auto tail = [](auto xs) { return xs([](auto first, auto ...rest) { return list(rest...); }); }; auto length = [](auto xs) { return xs([](auto ...z) { return sizeof...(z); }); }; // etc... // then use it like auto three = length(list(1, '2', "3")); The idea is to use the capture of the lambda as a fast compiler-generated struct with the desired members. By passing a function to that lambda, we get an access to the unpacked representation of those members and we can then apply any fast algorithm on parameter packs. Drawbacks --------- 1. The type of that tuple is completely opaque. Hence, it is harder to debug, but I'm working on other stuff that should make this a moot point. 2. Since lambdas can't be made constexpr, this tuple can't either. However, there was some discussion about proposing constexpr lambdas at C++Now and it looked like there was a lot of support for the idea, so this might change. Benchmarks ---------- I ran a couple of benchmarks testing the compilation time and the results are self explanatory. The benchmark suite is at [2]. ## Create a tuple of n elements The test code looks like: #include <boost/tuple/tuple.hpp> int main() { auto go = boost::make_tuple(1, ..., n); } <a href="http://imgur.com/rKqmvk3"><img src="http://i.imgur.com/rKqmvk3.png"/></a> ## Access the last element of a tuple of n elements The test code looks like: #include <boost/tuple/tuple.hpp> int main() { auto go = boost::make_tuple(1, ..., n); auto last = boost::get<n-1>(go); } <a href="http://imgur.com/b4rWR4y"><img src="http://i.imgur.com/b4rWR4y.png"/></a> Also, for those following the development of Boost.Hana, it is going well and I am preparing for a followup in the near future. Stay tuned! Regards, Louis Dionne [1]: http://github.com/ldionne/hana/blob/master/include/boost/hana/list.hpp [2]: http://github.com/ldionne/benchcc/tree/master/benchmarks

9 22

[optional] Add Optional to BoostBook subset of Boost C++ Libraries
by Andrzej Krzemienski 20 Jun '14

20 Jun '14

Hi, I am looking at this documentation index: http://www.boost.org/doc/libs/develop/doc/html/index.html And I can see that Optional is not included. Could someone add it there, or is there any criterion why certain libraries don't make it in? Regards, &rzej

3 2

[interprocess] Ticket 9414 (32 process communicating with 64 process)
by Joseph Van Riper 20 Jun '14

20 Jun '14

I ran into a problem with message queues that, after a couple of days investigation, lead to finding this bug. I notice that the maintainer has labelled this bug as 'invalid'. I would like to convince the maintainer otherwise. My alternatives are less optimum. I am working on a project where, on a Windows machine, I am running a service that waits for messages on the message queue. On a 32-bit environment, I will not run into any problems. But in a 64-bit environment, I may have either 64-bit or 32-bit clients attempting to write to the message queue, due to the nature of my needs and the nature of the Windows operating system. The messages from the clients are not specific to the bit-ness of the client, but are sent to the server. The server records these messages to a database it maintains. The server then pulls messages from the database and posts it to a web page. As the code stands now, I will have to compile both a 32-bit and 64-bit version of the server to install on 64-bit WinOS boxes, maintaining two separate databases of similar information, which will post to the web site with no sense of co-ordination (and little hope of co-ordination since they can't talk to each other). Or, alternatively, I would have to create my own message queue that doesn't rely upon std::size_t, but perhaps uses boost::uint32_t or boost::uint64_t instead... but it seems silly to me to maintain another message queue when this one is very close to being able to do what I need. I should also point out that I need for this to work in Linux as well, although I don't expect Linux has the Windows annoyance of having to work with two different bit-wise operating systems on one box. At the very least, I'd like a dialog, so I can understand why this defect is considered 'invalid'. Convince me that it really is invalid, and I'll go away. Otherwise, I respectfully argue against it. - Trey

3 4

blincubator comments
by Sohail Somani 19 Jun '14

19 Jun '14

Hi Robert, First of all, great idea and thanks for doing it. The concepts are solid and the site will be a great help for users and authors alike. It appeared that it was not possible to add any reviews anonymously. Is this by design? Does one have to register? It would be clearer if the page was unavailable or that there was a placeholder requiring the visitor to register before they could view the "submit review" page[1]. I closed the page after writing a review and being unable to submit because it seemed the site was buggy and so I set out to submit a report, but maybe the mistake was that I was supposed to have registered first. It was unclear. Also, if you ever do consider writing another version, consider Django because the admin interface[2] could allow you to generate a nice interface without futzing around with plugins and can also help you create a more maintainable site because it forces you into a structure. I'd be happy to help if you choose to go this way. General question: are there any other sites/applications of this kind out there? It seems to be a generally useful thing to make it easier to solicit user contributions for open source software. The closest would be a fork+pull request in Github that people can comment on. Maybe that's enough, I don't know. [1] http://rrsd.com/blincubator.com/review/?library_post_id=769 [2] http://www.djangobook.com/en/2.0/chapter06.html Sohail

3 4

[interest] Is there interest in planar angle library ?
by Matus Chochlik 19 Jun '14

19 Jun '14

On Tue, Jun 17, 2014 at 11:52 PM, Robert Ramey <ramey(a)rrsd.com> wrote: > how would this differ from boost units? > For one thing, in its dependencies, Boost.Units is a wonderful library, but for many use-cases it is too heavyweight. As mentioned before this library depends only on standard cmath. Also this library is more specialized on angle operations than Boost.Units is, at least out of the box. Matus

2 2

[gil] Optimizations
by Harg Tholan 19 Jun '14

19 Jun '14

Hi all, I've been wondering whether is there any room for ASM optimizations in GIL. Regards, H4rg.

2 1

[modularization] spirtit -> serialization
by Vicente J. Botet Escriba 19 Jun '14

19 Jun '14

Hi, comments about this dependency: |<boost/serialization/vector.hpp>| * from |<boost/spirit/home/support/detail/lexer/serialise.hpp>| It seems that this file is not used |boost/spirit/home/support/detail/lexer/serialise.hpp Could this file be removed or moves to examples? Best, Vicente |

10 58

Interest in bit-wise flags?
by Jeffrey Bush 19 Jun '14

19 Jun '14

Hi Boost! I have developed a type-safe bit-wise flags class for my own personal use and am wondering if it would be of interest for me to submit it as a Boost library. At the moment it is not ready for submission (code is messy) but I am testing the waters so I present an overview below. * Fully ISO standard C++11 (except for an MSVC version which is API-compatible except it doesn't support as many an infinite number of flag names) * They are type-safe, requiring explicit casts to and from integers/booleans * Supports standard bit-wise operations, and always keep values within the possible set of flags * Supports getting the name of a flag and looking up flags by name at run-time (when a flag is actually multiple-flags it gives something like "A|B") * Supports getting an array of all flags at run-time Now the bad things, which may be a deal-breaker: * Because of limitations in C++ the syntax can feel a bit awkward. You have to use a macro to define flag-names (which actually creates a class) then you can pass those flag-names as template arguments to the flags. An example: FLAG_NAME(A); FLAG_NAME(B); FLAG_NAME(C); typedef Flags<std::uint8_t, Flag_A, Flag_B, Flag_C>::WithValues<0x01, 0x02, 0x04> MyFlags; * I also couldn't get around using the ::WithValues nested class * I use a set of macros to make this cleaner (e.g. the above could be FLAGS(A, B, C)) and once defined there is no more ugliness Thanks for consideration! Jeff

5 5