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

Re: [Boost-users] Meta Template Programming with MPL - Compiler B ug or is the code really so evil slow?
by James Hughes 13 Feb '06

13 Feb '06

> -----Original Message----- > From: Vinzenz 'evilissimo' Feenstra [mailto:evilissimo@web.de] > Sent: 13 February 2006 12:40 > To: boost-users(a)lists.boost.org > Subject: [Boost-users] Meta Template Programming with MPL - > Compiler Bug > or is the code really so evil slow? > > > > Hi, > > I've written a little Meta Template Program which should > calculate the > first 3 perfect numbers. ( range 1 - 500 ) > > The code is compilable if you use the range 1 - 30 really > fast and works > correctly. But I've tried to compile the code but I can't get > it finished. > > Someone tried it on a Machine with 40 GB RAM and it now compiles for > more than 12 hours and uses more than 7.1 GB RAM ( the compilation ) > > I've tried it by myself with g++ 3.4.x and g++ 4.0.x and with VC++ 7.1 > > The code works on all tested compilers if the range would be > 1 - 30 but > > If the range is 1 - 500 VC++ says: > > d:\boost\vc71\include\boost-1_33_1\boost\mpl\aux_\preprocessed > \plain\apply_wrap.hpp(48): > fatal error C1204: compiler limit: internal structure overflow. > > ( This seems to be a known bug in VC++ > http://support.microsoft.com/default.aspx?scid=kb;en-us;883655 ) > > and g++ compiles and compiles and uses more and more RAM. > > > Is the code really so evil slow or is it a compiler Bug in g++? > > Maybe anyone has an answer. > > BR > evilissimo > > Without the code its difficult to tell, but this meta programming uses recursive definitions of templates, and with a lot of information required per template per iteration, you are running out of memory, as the error messages indicate. Not a compiler bug I don't expect - you are just asking too much. Templates require quite a bit of memory for a single instance - multiply that by the 500 you are asking for and things rapidly go downhill, which is why 30 works - you haven't hit the point at which you run out of memory (and hence require swapping which is a very slow process in comparison to memory access). In fact the multiplication factor is probably more than 500, it may go up as exponential or perhaps even factorial depending on the code James

2 1

Re: [Boost-users] Meta Template Programming with MPL - Compiler B ug or is the code really so evil slow?
by James Hughes 13 Feb '06

13 Feb '06

> -----Original Message----- > From: Vinzenz 'evilissimo' Feenstra [mailto:evilissimo@web.de] > Sent: 13 February 2006 12:40 > To: boost-users(a)lists.boost.org > Subject: [Boost-users] Meta Template Programming with MPL - > Compiler Bug > or is the code really so evil slow? > > > > Hi, > > I've written a little Meta Template Program which should > calculate the > first 3 perfect numbers. ( range 1 - 500 ) > > The code is compilable if you use the range 1 - 30 really > fast and works > correctly. But I've tried to compile the code but I can't get > it finished. > > Someone tried it on a Machine with 40 GB RAM and it now compiles for > more than 12 hours and uses more than 7.1 GB RAM ( the compilation ) > > I've tried it by myself with g++ 3.4.x and g++ 4.0.x and with VC++ 7.1 > > The code works on all tested compilers if the range would be > 1 - 30 but > > If the range is 1 - 500 VC++ says: > > d:\boost\vc71\include\boost-1_33_1\boost\mpl\aux_\preprocessed > \plain\apply_wrap.hpp(48): > fatal error C1204: compiler limit: internal structure overflow. > > ( This seems to be a known bug in VC++ > http://support.microsoft.com/default.aspx?scid=kb;en-us;883655 ) > > and g++ compiles and compiles and uses more and more RAM. > > > Is the code really so evil slow or is it a compiler Bug in g++? > > Maybe anyone has an answer. > > BR > evilissimo > > > Apologies wrt saying code snippet ws not present - have got it now, but comments on recursion still apply. James

1 0

Re: [Boost-users] Boost-users Digest, Vol 815, Issue 1
by Scott 13 Feb '06

13 Feb '06

> Robert Vinluan wrote: > > >Right now, in order to compile and link to the iostreams > library with > >zlib/bzip2 support, I have to include the .cpp files in my > project (as > >mentioned in the boost docs). I'm not too fond of this idea > as I want > >to keep 3rd-party code outside of my project build. > > > >Perhaps I have missed something in the docs, but is there > another way to > >link with the iostreams library without including the .cpp files? > > > > > Huh ? I don't understand... forgive me for my ignorance, since I'm > probably missing your point, but I just, for example, #include > <boost/iostreams/stream.hpp> and it just works, without > linking to any > special lib... what exactly are your needs that require > linking against > a lib? Or is it that bzip2/zlib support needs linking with a boost > library ? > > Regards, > > Leon Mergen > http://www.solatis.com/ I have the exact same problem Robert describes. I have compiled iostreams on Windows XP with zlib support (using my downloaded copy of zlib) and I added the various ZLIB parameters to bjam for compiling iostreams with zlib support. However, anytime I try to compile/link a program using iostreams with zlib, I get the same error message. (It's looking for a non-existent zlib library to link to, instead of the zlib library I compiled boost with).

2 1

Re: [Boost-users] progress_timer writes two newlines?
by Daniel Lidström 13 Feb '06

13 Feb '06

Beman Dawes scribbled: > Daniel Lidström scribbled: > > > > Hello! > > > > > > Why does boost::progress_timer write two newlines on destruction? > > > One would suffice for me. > > > Another thing about progress_timer: Why is the only option to > > output seconds? I'd like to be able to use my own formatter, so > > that I can have the output in HH:MM:SS.mmm for example: > > > boost::progress_timer timer(cout, hhmmssmmm_formatter()); > > Care to submit a patch? Here is a patch. With it, it is possible to use progress_timer as such: #include <boost/progress.hpp> #include <iomanip> using namespace std; // Time: S.mmm struct my_formatter { string operator()(double s) const { stringstream stream; stream << "Time: " << fixed << setprecision(3) << s << " s\n" ; return stream.str(); } }; // formats as HH:MM:SS.mmm string form(double s) { stringstream stream; int h = static_cast<int>(s/3600); int min = static_cast<int>((s-h*3600))/60; double sec = s - h*3600 - min*60; stream << setfill('0') << setw(2) << h; stream << ':'; stream << setfill('0') << setw(2) << min; stream << ':'; stream << fixed << setprecision(3) << setfill('0') << setw(6) << sec; stream << '\n'; return stream.str(); } int main() { { boost::progress_timer t(cout, my_formatter()); } { boost::progress_timer t(cout, form); } } Hälsningar, Daniel

1 0

[filestreams] Does native not work correctly in Windows?
by Paul Giaccone 13 Feb '06

13 Feb '06

The following line in my Windows program: boost::filesystem::path my_path(my_filename, boost::filesystem::native); converts backslashes in my_filename to forward slashes, which means than when I use boost::filesystem::exists on the result, the file is reported as not existing even though it does. I notice the comments in boost::filesystem::m_path_append(), which is called by the constructor of boost::filesystem::path, say that this is what the function does, but this contradicts the documentation. http://boost.org/libs/filesystem/doc/portability_guide.htm says, of native: Side effect: Syntax for path constructor /src/ string is implementation defined according to the path syntax rules for the operating system. Use: In path constructors, when the source is operator input or other sources which are formatted according to operating system rules. I understand this to mean that boost::filesystem::native converts the file into the format required by the OS. I certainly would not expect it to make the format wrong if it is already correct. Is there something I've overlooked here, or is this a bug, or is the documentation wrong (or just misleading)? I'm puzzled. Paul

2 3

Re: [Boost-users] Getting started with Boost: what did you need to know?
by Toby Smith 13 Feb '06

13 Feb '06

> I'm giving a talk about getting started with Boost > (https://www.cmpevents.com/SDw6/a.asp?option=G&V=3&id=271893) and I > thought it would be a good idea to solicit input from those who have > recently gone through the process. What do you wish someone had > explained to you? With what did you need hand-holding? Based on this > feedback I expect to make some improvements to the Boost website, > also. Any information you can give me will be much appreciated. Personally, my biggest problem with getting started with boost was realizing that I should get started in the first place. I'd had friends & coworkers mention boost, but to the new visitor, the website (as clean and readable as it is) doesn't do a good sell job for the product. It starts off well, saying that there's some sort of libraries to be had in here somewhere. But the first eight links the reader encounters have nothing to do with explaining why it is you might want to use these libraries. The ninth link, "Background information page," finally gives us a teaser, claiming our productivity will increase, but we still don't know why. Only the intrepid explorer who delves into the tiny "Documentation" link on the main page will be rewarded with a catalog of functionality. In short, I think the main page, the page that should be grabbing the casual surfer's attention, should at least highlight some of the functionality the consumer can expect from boost (cross-platform serialization, anyone? How about containers done right, or a well- designed thread package?), instead of devoting real estate and reader attention to links to random unix distros. Cheers, tob

4 3

[regex] - bk_parens and bk_braces missing from 1.33.x
by Keith MacDonald 13 Feb '06

13 Feb '06

Resending due to no response since 4th Feb: It states in history.html for regex 1.33.0 that "some of the more esoteric options have now been removed, so there is the possibility that existing code may fail to compile: however equivalent functionality should still be available". I suppose bk_parens and bk_braces could be considered esoteric, but I cannot find any equivalent functionality. Is that intentional? Thanks, Keith MacDonald

4 7

[ptr_container] ptr_map predicates
by Bill Buklis 13 Feb '06

13 Feb '06

How would I accomplish the following using ptr_map instead of std::map? typedef std::map< std::string, int* > stlmap; bool predicate( stlmap::value_type& mapitem ) { // return true if key value has at least 4 characters return( mapitem.first.size() > 3 ); } void sample( stlmap& themap ) { std::find_if( themap.begin(), themap.end(), predicate ); } The predicate for std::find_if requires *iterator for the parameter. However, with ptr_map *iterator returns a reference to the element value instead of the key/value pair. Normally this is fine (in general code I like this feature), but in this case I need access to the key in the predicate. How do I get at it? ptr_map version of above: typedef boost::ptr_map< std::string, int > bpmap; bool predicate( ??? ) { ??? } void sample( bpmap& themap ) { std::find_if( ??? ); }

3 12

boost::xtime quesion
by Tim Blechmann 13 Feb '06

13 Feb '06

hi all, maybe it's just a stupid newbie question, but i'm a bit confused by boosts timed locks ... they're using a time, that's based on the timeval struct on posix systems ... what's the difference between this and ptime from the date_time library? ptime can be used with a clock with microsecond granularity, too ... in any case, how do i convert times from ptime to xtime? thanks in advance ... tim -- mailto:TimBlechmann@gmx.de ICQ: 96771783 http://www.mokabar.tk Every word is like an unnecessary stain on silence and nothingness Samuel Beckett

1 0

[serialization] I am unable to link the serialization demo.cpp
by Michael Shapiro 12 Feb '06

12 Feb '06

I would like to use the boost serialization library to do checkpointing for a simulation which runs for one or more weeks. However, I have not been able to get past the "hello world" stage. I have tried to do this on both a linux machine running gcc 3.4.5 and Mac OS X using gcc 3.3. In both cases, my attempts to compile the demo.cpp program fail with long lists of undefined symbols. I will give the specifics for the Mac because this is where it will ultimately be used and because the instalation seems more standard there. The demo.cpp I am trying to compile is the one available on http://www.boost.org/libs/serialization/doc/tutorial.html The boost include files are installed in /usr/local/boost/gcc/ The libraries are installed in /usr/local/lib/ For example, here I find libboost_serialization.a and libboost_serialization.dylib. I am attempting to compile as follows: gcc \ -L/usr/local/lib/ \ -I/usr/local/include/ \ -lboost_serialization \ demo.cpp I have tried various variations on this including adding all the .a files to the list of libraries to link, and adding stdc++. I get error messages of which the following are typical ld: Undefined symbols: __ZNKSs5c_strEv __ZNKSt9basic_iosIcSt11char_traitsIcEE4failEv __ZNSaIcEC1Ev __ZNSaIcED1Ev followed by more of the same. I have no doubt there is something very simple I am failing to do, very likely something so simple no one knowledgeable would think to put it in the documentation. So I will probably feel very dumb when someone educates me about this, but I will also feel very relieved. Thanks in advance for your help.

2 1