April 2005 - Boost-users - lists.preview.boost.org

Compilers compliance and boost.
by Mauricio Gomes 20 Apr '05

20 Apr '05

According to these reports it seems there is not a single compiler able to compile 100% of boost, is that right ? http://www.meta-comm.com/engineering/boost-regression/cvs-head/user/ summary.html http://www.meta-comm.com/engineering/boost-regression/cvs-head/ developer/summary_release.html What is the best compiler to use boost ? Or in a less subjective way, what is the compiler that pass most tests ? Although it is mentioned here (http://www.boost.org/status/compiler_status.html) otherwise the link to the "Compiler Status Summary" page is only showing reports on Linux and it is not mentioning the version (CVS-HEAD ?). The other reports (links above) show the results for several platforms but they seem to be only available for CVS-HEAD. Where can I find the reports for a specific version like 1.32 for example ? Thank you for your attention, Mauricio Gomes Pensar Digital phone: 55-11-4121-6287 mobile: 55-11-8319-9610 http://pensardigital.com

4 8

loop vs. for_each & boost::bind
by Mathieu De Zutter 20 Apr '05

20 Apr '05

Hi, I have some code that runs fine: for (Partition::iterator pi = renderRegions.begin(); pi != renderRegions.end(); ++pi); (*pi)->render(&ltt, foo, si); I try to replace it by for_each & boost::bind, std::for_each(renderRegions.begin(), renderRegions.end(), boost::bind(&Region::render, _1, &ltt, foo, si)); but si doesn't get modified, while it should be passed as reference. Some more relevant code: class Region { ... void render(const LightTransportType* ltt, WorkAppraiser& wa, image::SampleImage& result); ... }; typedef std::vector< Region* > Partition; Partition renderRegions; Is this normal behaviour? -- With kind regards, Mathieu

4 3

RE: [Boost-users] Re: why so many .so and .a files for boost?
by Tom Tan (SH/RDC) 20 Apr '05

20 Apr '05

That 's right. Try command: $ls -l libboost_regex-gcc-d.so -rwxr-xr-x 2 tomtan staff 5.5M Feb 21 10:54 libboost_regex-gcc-d.so Now see the second column. more than 1 (2 in this case) means this file has hard links. Anyway, why not a symbolic link ? -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] Sent: Wednesday, April 20, 2005 17:14 To: boost-users(a)lists.boost.org Subject: [Boost-users] Re: why so many .so and .a files for boost? Sliwa, Przemyslaw (London) wrote: > I was wondering why I get duplicates on my Linux box as well. > The files are with 1_32 and without extension and they are exactly the > same. On any Unix I would expect those to be hardlinks to each other. Are you sure this is not the case? Markus _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

3 3

RE: [Boost-users] Re: why so many .so and .a files for boost?
by Tom Tan (SH/RDC) 20 Apr '05

20 Apr '05

got it, then why do i have to turn on the '-pthread' option of g++ when using boost_thread library? -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] Sent: Wednesday, April 20, 2005 14:51 To: boost-users(a)lists.boost.org Subject: [Boost-users] Re: why so many .so and .a files for boost? Tom Tan (SH/RDC) wrote: > Hi, all, > Why are there so many .so and .a files in $BOOT_HOME/lib? Take boost_singal as an example, there are about a dozen . They are: > libboost_signals-gcc-1_32.a > libboost_signals-gcc-1_32.so > libboost_signals-gcc.a > libboost_signals-gcc.so > libboost_signals-gcc-d-1_32.a > libboost_signals-gcc-d-1_32.so > libboost_signals-gcc-mt.a > libboost_signals-gcc-mt-d-1_32.so > .... > I wonder what are 'd' and 'mt' mean? "debug" and "mutil-thread"? If so, i wonder when i use a 'mt' thread , e.g. libboost_signals-gcc-mt-d-1_32.so, will i have to use -pthread for Linux and -pthreads for Solaris along with my other g++ compile options. Why? > Another, for stdc++.so, are there two versions , each for 'mutil-thread' and 'sigal-thread' support? What's the relationship between these versions and the compiler options like '-pthread' ? Answers at.. http://www.boost.org/more/getting_started.html#Results -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim - Grafik/jabber.org _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

2 1

wrong extension of shared libraries on HP/UX
by Oliver Kowalke 20 Apr '05

20 Apr '05

Hi, I've compiled boost on HP/UX with gcc. The extension of the shared libraries is *.so. I think it should be *.sl ! how can I fix this? thx, Oliver

1 0

wrong extension for shared libraries on HP/UX
by Oliver Kowalke 20 Apr '05

20 Apr '05

Hello, I've compiled boost on HP-UX with gcc-3.4.2. The shared libraries ends with *.so. On HP/UX the extension for shared libraries should be *.sl. So - is it a bug or what went wrong? thx, Oliver -- +++ GMX - Die erste Adresse f�r Mail, Message, More +++ 1 GB Mailbox bereits in GMX FreeMail http://www.gmx.net/de/go/mail

1 0

wrapping an mmap'ed pointer
by Davi de Castro Reis 20 Apr '05

20 Apr '05

Hi all, I am writing a class that returns some pointer to buffers in a mmaped area. Since this buffers have constant size, I am now wrapping them in a std::pair<const uint32, char *buf>. This works fine, but I can bet that boost have something better, with correct semantics for operator * and, maybe, ownership semantics. Another problem that I have with std::pair<const uint32, char *buf> is that I don't get the correct cv-qualified semantics with a const std::pair<const uint32, char *buf>. I would like the const modified version of my type to behave like a const std::pair<const uint32, const char * const buf>. The reason for this is that I will, then, be able to receive const char * buffers in my input. Check sample code below: class A { typedef std::pair<const uint32, char *> value_type; void put(const value_type buf); value_type get(); const value_type get() const; } Buffers in the input (put method params) will be copied inside my data structure and their memory management is up to the user. Buffers in the output (get method return values) are pointers to a mmap area and should not be freed. So, if I do: A a; a.put(make_pair(strlen("test"), "test")); Code doesn't compile because cv-qualification of the char * member of buf does not get correct. The only way to get out is to write this rather ugly class A. class A { typedef std::pair<const uint32, char *> value_type; typedef const std::pair<const uint32, const char * const> const_value_type; void put(const_value_type buf); value_type get(); const_value_type get() const; } Is there a better solution for this? boost::array doesn't fit the job because the size parameter is compile time. Smart pointers, besides missing the size member, calls delete() when it is not desired. So, a class to fit the job needs to have the size member (if there is no such class, it is ok to stick to std::pair or boost::tuple), and expected cv-qualification for the char *, meaning that neither the pointer nor the pointed-by area can be modified (btw, this problem arises often when I am working with boost::mpl, so a help here will be greatly appreciated). Thanks in advance, []s Davi de Castro Reis

1 0

aliasing make_tuple
by Davi de Castro Reis 20 Apr '05

20 Apr '05

Hi, I have a class where I use a tuple as input for some of my methods. The reason for this is that the number of parameters for these methods is a compile-time variable. See sample code below. template <uchar nparams, typename T1, typename T2 = none, typename T3 = none ...> class A { void f(boost::tuple<char, char, ..., char>); } So users can do something like A a<3, char, char, char> a.f(make_tuple('a', 'b', 'c'); The problem is that make_tuple is not a nice name, since it does not reflect the semantics of the parameter. Suppose I would like to call it key, so I can call A a<3, char, char, char>; a.f(key('a', 'b', 'c')); So, in my context, everything gets much more clear. Does someone know an elegant way of doing this? As I can thing, the only way I can do it, is to create a templatized free function called key, for which I will need to write 10 overloads (the max number of nparams, and also boost tuple, if I am not mistaken). Besides being a rather ugly solution, I will need to pay two copies of my objects before sending them to the f() method (with make_tuple I pay only one copy, which is also suboptimal, but, anyways...). Using #define is not an option since key is a very common name (the class is in a namespace). Ugly names such as MY_LIBRARY_KEY() are not a very nice also. Well, any tips? Solutions involving boost::mpl would be the best, but I couldn't find none. []s Davi de Castro Reis

2 3

RE: [Boost-users] Re: why so many .so and .a files for boost?
by Sliwa, Przemyslaw (London) 20 Apr '05

20 Apr '05

I was wondering why I get duplicates on my Linux box as well. The files are with 1_32 and without extension and they are exactly the same. Full debug/release install takes about 0.5G. This is a lot!! Pshemek -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Russell Hind Sent: 20 April 2005 09:09 To: boost-users(a)lists.boost.org Subject: [Boost-users] Re: why so many .so and .a files for boost? Rene Rivera wrote: > >> libboost_signals-gcc-1_32.a >> libboost_signals-gcc-1_32.so >> libboost_signals-gcc.a >> libboost_signals-gcc.so > > 'mutil-thread' and 'sigal-thread' support? What's the relationship > between these versions and the compiler options like '-pthread' ? > > Answers at.. http://www.boost.org/more/getting_started.html#Results But why duplicates of each? On windows I get libboost_signals-bcb-1_32.dll and libboost_signals-bcb.dll Which are both the same. Why are there libraries without the version number in created? Cheers Russell _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -------------------------------------------------------- If you are not an intended recipient of this e-mail, please notify the sender, delete it and do not read, act upon, print, disclose, copy, retain or redistribute it. Click here for important additional terms relating to this e-mail. http://www.ml.com/email_terms/ --------------------------------------------------------

3 2

why so many .so and .a files for boost?
by Tom Tan (SH/RDC) 20 Apr '05

20 Apr '05

Hi, all, Why are there so many .so and .a files in $BOOT_HOME/lib? Take boost_singal as an example, there are about a dozen . They are: libboost_signals-gcc-1_32.a libboost_signals-gcc-1_32.so libboost_signals-gcc.a libboost_signals-gcc.so libboost_signals-gcc-d-1_32.a libboost_signals-gcc-d-1_32.so libboost_signals-gcc-mt.a libboost_signals-gcc-mt-d-1_32.so .... I wonder what are 'd' and 'mt' mean? "debug" and "mutil-thread"? If so, i wonder when i use a 'mt' thread , e.g. libboost_signals-gcc-mt-d-1_32.so, will i have to use -pthread for Linux and -pthreads for Solaris along with my other g++ compile options. Why? Another, for stdc++.so, are there two versions , each for 'mutil-thread' and 'sigal-thread' support? What's the relationship between these versions and the compiler options like '-pthread' ? BR. /TomTan

3 2