August 2006 - Boost - lists.preview.boost.org

[asio] new resolver query with numeric ports
by christopher baus 17 Aug '06

17 Aug '06

I'm converting some code from the previous asio release to the latest 0.37 and I am confused by the new resolver interface. ip::tcp::resolver resolver(io_service); ip::tcp::resolver::query query("foobar.com", "http"); ip::tcp::resolver::iterator iter = resolver.resolve(query); ip::tcp::resolver::iterator end; while (iter != end) { ip::tcp::endpoint ep = *iter; std::cout << iter->endpoint() << std::endl; std::cout << iter->host_name() << std::endl; std::cout << iter->service_name() << std::endl; ++iter; } The resolver::query requires a service name string. I want to specify a port number so that the endpoints returned by the resolver::iterator are configured with that specified port rather than those returned by the service name. It isn't clear to me, but it looks like it should be possible by passing numeric_service in the query flags, but on Windows AI_NUMERICSERV isn't defined and numeric_service == 0. Plus that leaks the rather ugly getaddrinfo() interface of using numeric strings. It would be nice if the interface allowed: ip::tcp::resolver::query query("foobar.com", 80); I feel like I might be missing something here. Thanks, Chris

2 2

Update on release plans and formal review schedule
by Richard Newman 17 Aug '06

17 Aug '06

Hello all, What is the current status of a next Boost release? Do we have any kind of an ETA (the last release was in December 2005)? I notice that the regression tests seem to settle in the low 200s and then spike up again every so often, so from the "outside" appears that we are not closing in realistically on a release. Are there still formal library reviews pending? The boost.org website does not appear up to date with their status. I'd like to see a release soon that rolls up inclusion of approved libraries, such as asio, and re-establish a baseline of bug fixes and improvements. Thank you. Kind regards, Richard Newman richard(a)cdres.com

3 3

[process] - suggestion/addition to posix_start
by Oliver.Kowalke＠qimonda.com 17 Aug '06

17 Aug '06

Hello, for posix_start I would suggest following in order to pass the error of ::execve to the parrent: int fd[2]; if ( ::pipe( fd) < 0) // create a pipe throw sys::system_error( errno, sys::errno_ec); if ( ( pid = ::fork() ) < 0) // fork new process throw sys::system_error( errno, sys::errno_ec); if ( pid == 0) // child { ::close( fd[0]); // close one end of the pipe ::fcntl( fd[1], F_SETFD, FD_CLOEXEC); // close other end of pipe if this process is replaced by ::execve struct sigaction act; // clear all signal handlers ::sigemptyset( & act.sa_mask); act.sa_handler = SIG_DFL; act.sa_flags = 0; for ( int sig = 1; sig < NSIG; sig++) ::sigaction( sig, & act, 0); ... ::execve( cl.get_executable().c_str(), args.second, envp); // replace by new application int msg = errno; // reached only if ::execve failed ::write( fd[1], & msg, sizeof msg); // write error number to pipe ::_exit( errno); // exit } // parent ::close( fd[1]); // close one end of pipe while( true) { int msg; // error number passed from child process int n = ::read( fd[0], & msg, sizeof msg); if ( n == sizeof( int) ) { // child has written something to pipe ::close( fd[0]); throw sys::system_error( msg, sys::errno_ec) ); } else if (n == -1) { if ( errno == EINTR) continue; } // interrupted break; // EOF of pipe } ::close( fd[0]); return child( pid); regards, Oliver

1 0

Re: [boost] Presentation of Boost.Process (SoC project)
by Oliver.Kowalke＠qimonda.com 17 Aug '06

17 Aug '06

Hi, I would suggest to implement the possiblility to wait for a child/children (non blocking) inside of a SIGCHLD signal handler function. On POSIX the implementation would look like: // while loop inside of signal handling function (SIGCHLD) while ( true) { pid_t pid; int stat = 0, opts = WNOHANG | WUNTRACED; #ifdef WCONTINUED opts |= WCONTINUED; #endif pid = ::waitpid( -1, & stat, opts); if ( pid == 0) break; else if ( pid < 0) throw sys::system_error( sys::error_code( errno, sys::errno_ec) ); child c( pid, status( stat) ); ... } regards, Oliver

3 2

[ Boost.Processt] - feature request
by Oliver.Kowalke＠qimonda.com 17 Aug '06

17 Aug '06

Hello, I suggest following features as additions to boost.process: - status should also check for continued state with WIFCONTINUED and return SIGCONT as signal - posix_child::wait should support the flags WUNTRACED and WCONTINUED - new satic function static std::pair< posix_child, status > posix_child::wait() which supports the flags WNOHANG, WUNTRACED and WCONTINUED (used inside of SIGCHLD signal handler) - posix_child shoulkd implement following functions exist -> ::kill( pid, 0), stopp -> ::kill( pid, SIGSTOP), continue -> ::kill( pid, SIGCONT), terminate -> ::kill( pid, SIGTERM), kill -> ::kill( pid, SIGKILL), usr1 -> ::kill( pid, SIGUSR1) and usr2 -> ::kill( pid, SIGUSR2) - posix_launcher should support rlimits -> ::set_relimit(), priority -> ::setpriority(), process group id -> ::setpgid(), session leader -> ::setsid(), creation mask -> ::umask() regards, Oliver

1 0

Request for a dynamic_bitset<> maintainer
by Gennaro Prota 17 Aug '06

17 Aug '06

Hi, according to <http://www.boost.org/more/submission_process.htm> I'm asking for someone to take over as the library maintainer of dynamic_bitset<>. Until August 14 I can help understanding the details if that's ever needed. Cheers, Gennaro. -- [ Gennaro Prota, C++ developer for hire ]

2 1

Re: [boost] Regression Analysis Algorithm
by Wang Weiwei 17 Aug '06

17 Aug '06

>| -----Original Message----- >| From: boost-bounces(a)lists.boost.org >| [mailto:boost-bounces@lists.boost.org] On Behalf Of Wang Weiwei >| Sent: 17 August 2006 09:33 >| To: boost(a)lists.boost.org >| Subject: [boost] Regression Analysis Algorithm >| >| Hello everyone, >| >| Is there any regression analysis algorithm/package in the >| boost librabies, or elsewhere? > >No - but some of necessary building blocks are available, or upcoming: > >UBLAS/LAPACK, Matrix handling, Eric Neibler's accumulator, John Maddock's >math & stats functions you need to do statistics (watch out for a preview >Real Soon Now!). > >So you may be able to write one for Boost before too long ;-) Thanks. :-) > >Paul > >--- >Paul A Bristow >Prizet Farmhouse, Kendal, Cumbria UK LA8 8AB >+44 1539561830 & SMS, Mobile +44 7714 330204 & SMS >pbristow(a)hetp.u-net.com >

1 0

Regression Analysis Algorithm
by Wang Weiwei 17 Aug '06

17 Aug '06

Hello everyone, Is there any regression analysis algorithm/package in the boost librabies, or elsewhere? Thanks. Max

2 1

Re: [boost] [boost.asio] file / pipe io service
by Christopher Kohlhoff 17 Aug '06

17 Aug '06

Hi Martin, Martin Slater <mslater(a)netspace.net.au> wrote: > > The read failed with WSAECONNRESET, which I take as "will > > not work". > > > > But I would be glad to hear, that it will work, and that I > > have to only find out what I did wrong. > > > Sorry I wasn't as clear as I could be, I'm only using sockets > for what I would like to use named pipes for. I use regular > win32 ReadFile in a thread currently to receive output on > stdout/err and was hoping that this would fit nicely into the > asio model too and let me remove this code and the neccessary > synchronisation.. I think the email to which you're replying must have gone astray -- at least I can't find it in my email client or in the list archive. Can you post the whole problem again, thanks. Cheers, Chris

2 2

Re: [boost] Boost Units library preview
by Eric Lemings 17 Aug '06

17 Aug '06

> -----Original Message----- > From: boost-bounces(a)lists.boost.org > [mailto:boost-bounces@lists.boost.org] On Behalf Of Andy Little > Sent: Wednesday, August 16, 2006 7:46 AM > To: boost(a)lists.boost.org > Subject: Re: [boost] Boost Units library preview > > Hi Eric, > ... > > I'm not quite sure which way to take this. Does this mean > that you consider the Quan UI unsatisfactory and beyond repair? Well I'll put it this way. Let's say you have two library interface designs. The first design, A, is the hypothetically perfect library interface. The second design, B, is a tangible, real, albeit less than perfect design. The goal of course is to get B as close as possible to A. Now would it be easier to hack away at B or start from scratch with a new design called C? I decided to take the latter approach. Compare the library interface of Quan to my proposed design. I would be interested to know what you think. > The implementation is loosely based on the dimensional > analysis example in David Abrahams and Alexei Gurtovoys TMP > book The implementation can be seen here: That's a start but it's not a straight jacket. In addition to seperating the concept of a 'quantity' from the concept of a 'measure', I wanted to allow the user to define dimensions and units in a free-form manner using a type sequence of base dimensions and units. The user can list the dimensions and units of the sequence in any order. > Tested in VC7.1 and VC8.0 of vthe two examples gave compile > errors. Testing in gcc 4.0 gave a failed assertion at runtime Like I said, it's a preview. I made it available in order to get comments and feedback regarding the design of the interface. It's not even close to being ready for real use. > Is the structure of a unit / conversion factor based on that > in quan?. It certainly looks very similar. Yes, same approach. (Only solution to the problem that has been devised AFAIK.) Eric.

2 1