- Boost-users - lists.boost.org

Using external property map with a mutable boost.graph
by Jeff Holle 30 Aug '04

30 Aug '04

I'm using a boost.graph via: typedef boost::adjacency_list<boost::vecS,boost::vecS,boost::bidirectionalS> GraphT; In my application, I've decided to use external maps. I have discovered a problem using vertex descriptors as a key of a map. Basically, in doing a remove_vertex sequence involving vertex descriptors 5, 3, 2, I discovered that the next vertex descriptor produced by add_vertex was 4. At this point the insert into my map failed because this element already existed. Given the mutable nature of vertex descriptors, I don't see how they can be used as a key in a map. I'm contemplating two options at this point. 1. Switch from map to vector in storing my vertex property externally. 2. Switch to internal properties. My questions are: 1. Is the first option viable? As I see things, vertex descriptors are always contiguous and in the range (0..num_vertices(graph)]. Is this true? 2. Do internal "maps" avoid this type of problem?

2 1

date_time bug cvs latest vc7.1
by Gordon Smith 29 Aug '04

29 Aug '04

The following sample will not compile with the following settings (I know it has been discussed with regards to the regex library previously) - this code compiles fine in 1_31: "Character set": "Use Multi-Byte Character Set" (I am targetting win 95/98) "Treat wchar_t as Built-in Type (/Zc:wchar_t)": "No" The contrived example demonstrates why I can't turn "/Zc:wchar_t" on: #include "stdafx.h" #include <comutil.h> #include <boost/date_time/gregorian/gregorian.hpp> int _tmain(int argc, _TCHAR* argv[]) { _variant_t var("Narrow String"); return 0; } begin 666 boost_bug_001.zip M4$L#!!0````(`&A9$C&80G7MR@````\!```1````8F]O<W1?8G5G7S P,2YC M<'!5CD%+Q# 0A<\&\A^&>MD52>K5E071@R</ZDTD9-,T'6@G83I=%?&_V^Q% MO7WO8^;QK(5#SK.XPY)<VUZ94 I<PWWLD>(,,D2()/P))2,)])E/+F2:\QC! MES)B\(*9C%;6:J75.5(8ERY",TOG^P\S-/_T3<C3(CB:8?_7GF;8SDMT@E.T MB6/*C)Y^R0RE[&M7G>)D\DB;BIY3N 3W<O=P^W11T_'U;:O5EU9G[NCKISB! ME3;-HV?.[_ LC)2:[6X]X2@+$[0K?_\`4$L#!!0````(`%!6$C$O3O0P5P$` M`),#```1````8F]O<W1?8G5G7S P,2YS;&Z=D,UKPC 8QL\6^C^$GAR8DFCM MZF&'IA]CA\%0V([2CR(a)962)-*AO._WVI5ITZF'H(29XW[^]]\CRSHI)*SC1X M9:K..)CHNF023"2O-9,"I(S3'DAE]9&9-[12C1BX"-G62R7?::&[SBH@T2A* M2 `#XH40XQC!(,88(A2B:(1-U>NOG3OP`)Q<2J6G>3V?(H2=WHG@+HN%H3;Z M:NCA*"+W'D1ID$+/1Q@&89I GY!1WP_]88(':\>V.JV/B5G&7+>]QG1!14E% MP:AJ)B_,F+9D>A)1'K?9UD&RK4<N\XR;=]O##KU+)9)BQN9UE367#;RBNYII MZL34_,?(F[T1QI333%$CM:>MA2/ZV;36S?DP\Y/?TRY+RMV8<4/#7M)HMK?W M_<;$H'\UA]2,ERZZF=+F<.2GU6YF'3R=DO[-.OG45"B6,\[TU[:D_LCZ.DY8 MED_B,LQ>L:T?4$L#!!0````(`%!6$C&SGO.4R0```"P!```*````<W1D869X M+F-P<&60O6[#, R$=P-^AT,ZUT[7; 6"HEN&9.@6T!)=*U!(03]H_/:17:-+ MUR._XQW['BE;&A^="0$')"W1,$;G&7FB#"?&%\L)MY)RE;CNDUB*]F_4-GV/ M037EZU"^K_O]6Q?,A!_G/09>F1#YU>@]5%N+B<ER7*GMM@ZWWW6CDLG)?R;/ M@>O!4>.=LE-IF[9YV0)@M]E,NT6NMI?3\71 Y)$C2ZU#,H.L=0M)?@N0,&N! M,"]-<+X<WS^^NL^5KP4ANI2O25Q:W]$V3U!+`P04````" !05A(QJ,6F9;L` M```E`0``" ```'-T9&%F>"YH58_!BL,P#$3O@?S#P%Y+<R]+3WOO97] :\NQ M2V)G91F:OZ]<*&TO`FF>1J-I0E5/X7:,."%EMS3/"&FQ4L0TRI[$H^Y5>?T` MZF$<I@E&;5*N[!1U8Y="<I\8-)*"A-$J>P3A_\99E_V OZ8/CRZZ2'DV/>47 MT<5Q&(>O36A>"24[[KU-GB>^4ZDJ3.OY?:CF)L=X[JCY_UY^+B<(!Q8V"Y#W M25/)M" R>9:*O;3'([/0BAX@B46/MC .=U!+`0(4"Q0````(`&A9$C&80G7M MR@````\!```1``````````$`( ````````!B;V]S=%]B=6=?,# Q+F-P<%!+ M`0(4"Q0````(`%!6$C$O3O0P5P$``),#```1``````````$`( ```/D```!B M;V]S=%]B=6=?,# Q+G-L;E!+`0(4"Q0````(`%!6$C&SGO.4R0```"P!```* M``````````$`( ```'\"``!S=&1A9G@N8W!P4$L!`A0+% ````@`4%82,:C% MIF6[````)0$```@``````````0`@````< ,``'-T9&%F>"YH4$L%!@`````$ -``0`[ ```%$$```````` ` end

2 3

boost::thread callback copy problem
by Howard Cole 27 Aug '04

27 Aug '04

Hi, I am new to boost::thread but have come across the following problem and wondered if anyone could help... The problem is the copy overhead used by the boost::thread / boost::function0 code. The following code actually creates multiple copies of myclass rather than using a reference to myclass and due to the size of the class, this leads to a considerable overhead. I only noticed this problem because the thread kept failing due to the fact that my constructor held some socket initialisation code, which was being called multiple times. Is there a way that I can force the thread class to use references rather than copying? Looking at the code for the function library, I see that you can use boost::ref for function classes, any ideas as to how it can be used in the thread class? Thanks in advance Howard Cole www.selestial.com Example code. class myclass { public: // Lots of local data here void operator() () { do_thread_stuff(); }; // CTOR myclass(){ do_some_initialisation_stuff(); }; // DTOR ~myclass(){ do_some_cleanup_stuff(); }; // Lots of member funcs ... }; void main (void){ myclass test_instance; boost::thread test_thread(test_instance); test_thread.join(); }

3 3

RE: [Boost-users] Re: Guidelines for effective exception usage
by Mark Storer 26 Aug '04

26 Aug '04

> No it is not. Even if you aren't throwing due to memory starvation, > you could run out of memory during unwinding, when the exception is > copied. That leads you directly to terminate(). Do not pass Go; do > not collect $200. Venturing a little far afield, but one technique I've heard of is to grab a block, so you can release it in low-memory situations (or just every time you throw). This ensures (in theory) that you'll have enough memory to copy your exception, create your strings, and so on. If you manage to recover, you just grab another block ("prime the pump") and soldier bravely forward. Given STL's habit of allowing people to specify an allocator, you could even go so far as to build a custom allocator that worked on some pre-allocated block of memory for your exceptions (strings, vectors, whatever you deem necessary). Either way, you'll run out of memory a little earlier (not a big issue in many environments), but you _will_ get to pocket the $200. It's not hard to imagine an environment where grabbing a couple extra Kb would be a Bad Thing, but I'd think this trick would be fine on PCs and the like. --Mark Storer Senior Software Engineer Verity, Inc.

4 3

Generative Programming book
by Samuel 26 Aug '04

26 Aug '04

The ISBN for the "Generative Programming" book is wrong. 0-210-30977-7 is incorrect 0-201-30977-7 is correect That book is used and described in the Boost "Feature Model Diagrams in text and HTML" page. Also, the Generative Programming web site isn't working for me, but perhaps the problem is only temporary. Note that I am not asking for help with any of that. I obviously already know what the correct ISBN is. I am posting this in case it helps others; the web page needs to be updated for the ISBN at least. The Boost "What to do about Boost bugs" page says to submit "a bug report to either boost-users list, boost mailing list, or our bug tracking facility at SourceForge" and that "either of the mailing lists is a preferred way". However it does not describe what a bug report is. If there is not a formal format, then I hope the above is adequate.

2 1

/usr/include/boost/mpl/if.hpp:75: internal compiler error: Segmentation fault
by Nils Wagner 26 Aug '04

26 Aug '04

Dear experts, I am going to install VPython http://www.vpython.org/linux_download.html on SuSE 9.1 which makes use of the boost library. However I got a segmentation fault. /usr/include/boost/mpl/if.hpp:75: internal compiler error: Segmentation fault Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://www.suse.de/feedback> for instructions. Preprocessed source stored into /tmp/ccrXaXyw.out file, please attach this to your bugreport. Any pointer would be appreciated. Nils

2 2

Boost Build, custom-built targets
by John Meinel 26 Aug '04

26 Aug '04

Hello all- I'm currently using Boost Build v1 for some of my projects. I must say, it's significantly better than Makefiles. Overall, I'm quite pleased, though it took a little while to learn. I haven't been able to figure out one thing, though. How to generate custom built targets. Basically, I want to auto-generate a C++ source file so that I can keep track of version information. I have a python script that determines the version information from various sources (such as the revision control system [arch]), and I would like it to be run before my library is built. If I was using a makefile I would do something like this: .PHONY: version.cpp version.cpp: generate_version.py version ./generate_version.py This would cause version.cpp to be generated fresh each time the build process was called. However, I don't really need the ".PHONY", it would be okay if it was only created whenever one of the other files changed. I'm sure it's possible to do this with bjam, I just haven't figured out how to do customized stuff like this yet. Thanks, John =:-> PS> Is there a better mailing list for this?

2 1

RE: [Boost-users] Re: Guidelines for effective exception usage
by Ben Hutchings 25 Aug '04

25 Aug '04

Peter Dimov <pdimov(a)mmltd.net> wrote: > Ben Hutchings wrote: > > Peter Dimov <pdimov(a)mmltd.net> wrote: > >> Victor A. Wagner Jr. wrote: > >>> > >>> OK, blunt point. > >>> the standard is (insert favorite expletive or "in error") if it > >>> allows use of std::runtime_error to terminate the program due to > >>> low memory situations (run out of memory (due to copying) during > >>> stack unwinding). > >> > >> It does not allow such a thing. > > > > This is an known and unresolved issue: > > http://www.open-std.org/jtc1/sc22/wg21/docs/lwg-active.html#254 > > "The copy constructors of all exception-derived classes already have > a no throw spec. Reference 18.6.1, 19.1 and 15.4/13." "As they stand, the copy constructors might fail via a call to unexpected. I think what is intended here is that the copy constructors can't fail." (Is it really necessary for us to quote the referenced text at each other? This is silly.)

3 2

RE: [Boost-users] Re: Guidelines for effective exception usage
by Ben Hutchings 25 Aug '04

25 Aug '04

Peter Dimov <pdimov(a)mmltd.net> wrote: > Victor A. Wagner Jr. wrote: > > > > OK, blunt point. > > the standard is (insert favorite expletive or "in error") if it > > allows use of std::runtime_error to terminate the program due to low > > memory situations (run out of memory (due to copying) during stack > > unwinding). > > It does not allow such a thing. This is an known and unresolved issue: <http://www.open-std.org/jtc1/sc22/wg21/docs/lwg-active.html#254>

2 1

Update: exception class that
by Stephen torri 25 Aug '04

25 Aug '04

I thought I would share what I am using for exceptions given the excellent advice from everyone responding to my initial email. There is no usages of any memory allocation or std::string like variables. If there are any 'oops' or 'gotchas' please let me know. My plan is to use a command flag for taking into a value. The value will determine what debug messages are reported at the cause of the problem. The message will be displayed via std::cerr. Stephen ---------------- Header ---------------- #ifndef IO_EXCEPTION_H_ #define IO_EXCEPTION_H_ #include <exception> #include <inttypes.h> namespace kurt { /*! * \class IOException * \brief Rather than inheriting from ifstream this class delegates * any required functionality to it. * \date 2003 * \author Stephen Torri */ class IO_Exception : public virtual std::exception { public: IO_Exception (uint32_t message_id); /*! * \brief Destructor */ virtual ~IO_Exception (void) throw(){} /*! * \brief Return the extra message accompanying the exception. */ const char* what (void) const throw(); /*! * \brief Return tne name of the exception */ const char* name (void) const throw(); enum { CANNOT_OPEN, INPUT_CORRUPTED, INVALID_IO_OP, READ_ERROR, UNSUPPORTED_OP, WRITE_ERROR, }; private: static const char* m_exception_name; static const char* m_messages[]; uint32_t m_id; }; } /* namespace kurt */ ------------- Source ------------- #include "IOException.hpp" namespace kurt { const char* IO_Exception::m_exception_name = "IO_Exception"; const char* IO_Exception::m_messages [] = { "I/O Error: cannot open", // CANNOT_OPEN "Input data is corrupted", // INPUT_CORRUPTED "Invalid i/o operation attempted", // INVALID_IO_OP "Error reading input", // READ_ERROR "Unsupported I/O operation", // UNSUPPORTED_OP "Error writing output", // WRITE_ERROR }; IO_Exception::IO_Exception (uint32_t message_id) : std::exception(), m_id(message_id) {} const char* IO_Exception::what (void) const throw() { return m_messages[m_id]; } const char* IO_Exception::name (void) const throw() { return m_exception_name; } } /* namespace kurt */ #endif /* IO_EXCEPTION_H_ */ -- Email: storri(a)torri.org

1 0