November 2002 - Boost-users

Is intrusive_ptr the thing to use?
by Hickman, Greg 26 Nov '02

26 Nov '02

I need an object managed through a shared_ptr to be able to pass itself (i.e., this) to another object which also expects a shared_ptr. Is there a preferred implementation technique for this when using the boost smart pointers? I think I've seen conversation in the past about deriving from boost::counted_base and calling shared_from_this(), but I think I also remember seeing comments to the effect that this is going away. OTOH, I know there's something called intrusive_ptr that's currently undocumented. Should I be using one of these approaches? If the answer is intrusive_ptr, are there docs? Should I roll my own? What's a Booster to do? -Greg

3 2

matrix inside a matrix?
by Stephen Crowley 26 Nov '02

26 Nov '02

Is it possible to define a matrix that is part of a larger matrix, without having to create temporaries and copy part of the larger? /----------\ | | | | SM | | |----| LM | | | \----------/ I need to extract a sub-section of a matrix, multiply it, and then copy it back. I am currently using a temporary, but I have a feeling there must be a better way. -- Stephen

2 1

RE: [Boost-Users] Is intrusive_ptr the thing to use?
by Mark Storer 25 Nov '02

25 Nov '02

>The only problem with it currently is that the >smart_class MUST be managed with smart_ptrs, >if you try to 'delete' it, it will try to delete >itself and crash. This seems like a problem that could be overcome through overloading the "delete" operator. Perhaps if you declared your delete operator as protected (or private), and then made the smart pointers you wanted to use friends? That would force you to use a specific set of smart pointer classes, but that doesn't sound like it would be a problem in your case. It's fixing the symptom rather than the disease, but is better than nothing. Turning runtime errors into compiletime errors is always a Good Thing. --Mark Storer Software Engineer Cardiff Software #include <disclaimer> typedef std::disclaimer<Cardiff> Discard;

2 1

RE: [Boost-Users] Re: uBLAS, MTL performance tests.
by Randy Bowen 25 Nov '02

25 Nov '02

Actually, default behavior of the IDE is to define "_DEBUG" in the debug configurations, and "NDEBUG" in the release versions. Those are set as defaults for all files, but may be overridden on a case-by-case basis as necessary. This is a bit off topic for the Boost list. If you have other questions in this area, feel free to email me directly. Randy Bowen rbowen(a)stamps.com -----Original Message----- From: Jon Agiato [mailto:JonAgiato@nyc.rr.com] Sent: Friday, November 22, 2002 6:32 PM To: Boost-Users(a)yahoogroups.com Subject: Re: [Boost-Users] Re: uBLAS, MTL performance tests. Thanks Randy! Would you know if it is also set for release by default as well? Jon Agiato JonAgiato(a)nyc.rr.com ----- Original Message ----- From: Randy Bowen To: Boost-Users(a)yahoogroups.com Sent: Friday, November 22, 2002 6:40 PM Subject: RE: [Boost-Users] Re: uBLAS, MTL performance tests. > > > add -DNDEBUG to command line arguments. Unfortunately > I don't know where exactly you can edit command line > arguments in VC.NET. > > Alexei. > In VC.NET, this can be set under Project Properties -> C/C++ -> Preprocessor -> Preprocessor Definitions. (It's normally preset for debug configurations.) Randy Bowen Yahoo! Groups Sponsor ADVERTISEMENT Info: <http://www.boost.org> Wiki: <http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl> Unsubscribe: <mailto:boost-users-unsubscribe@yahoogroups.com> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service. [Non-text portions of this message have been removed] Info: <http://www.boost.org> Wiki: <http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl> Unsubscribe: <mailto:boost-users-unsubscribe@yahoogroups.com> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.

1 0

johnson algorithm, bug or feature?
by Ramón Casero Cañas 25 Nov '02

25 Nov '02

I have tested the johnson_all_pairs_shortest_paths example of libboost-graph 1.28. It's a directed graph. Shortest path from vertex 3 to vertex 0 weights "inf" (all edges of vertex 0 are outgoing edges). That's OK. Now we change the graph from directedS to undirectedS. The path weights now "5": 0 1 2 3 4 5 0 -> 0 0 -1 -5 0 -4 1 -> 0 0 -1 -5 0 -4 2 -> 1 1 0 -4 1 -3 3 -> 5 5 4 0 5 1 4 -> 0 0 -1 -5 0 -4 5 -> 4 4 3 -1 4 0 But if the path is 3-> 4 -> 0, should it not be "-5"? Why does the algorithm change the sign in an undirected graph? In this reference of NIST http://www.nist.gov/dads/HTML/johnsonsAlgorithm.html we can read: "Johnson's algorithm (algorithm) Definition: An algorithm to solve the all pairs shortest path problem in a sparse weighted, directed graph". So, is it really the problem that Johnson's algorithm doesn't work on undirected graphs? If so, shouldn't the function throw a compilation error/warning and/or be warned in the documentation? Regards, Ramón.

1 0

RE: [Boost-Users] Re: uBLAS, MTL performance tests.
by Randy Bowen 23 Nov '02

23 Nov '02

> > > add -DNDEBUG to command line arguments. Unfortunately > I don't know where exactly you can edit command line > arguments in VC.NET. > > Alexei. > In VC.NET, this can be set under Project Properties -> C/C++ -> Preprocessor -> Preprocessor Definitions. (It's normally preset for debug configurations.) Randy Bowen

2 1

RE: [Boost-Users] What is wrong with the GLPL?
by Tom Matelich 23 Nov '02

23 Nov '02

> From: Steve M. Robbins [mailto:steven.robbins@videotron.ca] > Sent: Friday, November 22, 2002 5:43 PM > On Fri, Nov 22, 2002 at 10:01:59AM -0500, > scleary(a)jerviswebb.com wrote: > > > There is a second problem with the LGPL in its restrictions > of modified > > copies of the original source (i.e., you can't take an LGPL > library, enhance > > it, and then sell it). > > I am unable to let this go unchallenged. > > There is not any provision in the license language that prevents one > from enhancing an LGPL library and selling it. > > Most likely you meant to say "you can't take an LGPL library, enhance > it, and then sell it while hiding your enhancements". That would > be a true statement. In the spirit of challenging... If you take an LGPL library, and enhance it, you must provide the source for the changes you made, so selling it wouldn't do you much good, that I can see. ----------------------------------------------------------------------- DISCLAIMER: Information contained in this message and/or attachment(s) may contain confidential information of Zetec, Inc. If you have received this transmission in error, please notify the sender by return email. -----------------------------------------------------------------------

1 0

RE: [Boost-Users] What is wrong with the GLPL?
by scleary＠jerviswebb.com 23 Nov '02

23 Nov '02

> -----Original Message----- > From: Markus Werle [mailto:yg-boost-users@m.gmane.org] > > Hi! > > I just came across http://www.boost.org/more/lib_guide.htm#License > where it is stated that any boost library license > > - Must be simple to read and understand. > - Must grant permission to copy, use and modify the software for > any use (commercial and non-commercial) for no fee. > - Must require that the license appear on all copies of the > software source code. > - Must not require that the license appear with executables > or other binary > uses of the library. > - Must not require that the source code be available for > execution or other > binary uses of the library. > - May restrict the use of the name and description of the library to > the standard version found on the Boost web site. > > At the same place one can find > > "Restricted licenses like the GPL and LGPL are not acceptable." > > I may have overlooked something, but I feel like the LGPL meets the > requirements above. Could You please explain or point to further > information about this issue. The main problem with LGPL is the "relinking" clause, which requires any distributer of a work that uses the library to allow the end-user to re-link the executable with a different version of the LGPL'ed library. While this makes sense for traditional libraries, it causes real problems when templates get involved; think of what this means for template libraries covered by the LGPL: in order to allow the user to relink with a newer version of the LGPL library (which is the intent of this clause), the source that uses the LGPL'ed templates must be distributed. There is a second problem with the LGPL in its restrictions of modified copies of the original source (i.e., you can't take an LGPL library, enhance it, and then sell it). This is not to imply that LGPL or GPL are bad choices for a project; I'm working on a GPL project myself. But they don't fall under Boost's "totally free" licencing guidelines. If you want more info for your licencing decision, check out the Open Source Initiative (http://www.opensource.org/) -Steve

2 1

uBLAS, MTL performance tests.
by alexei_novakov 22 Nov '02

22 Nov '02

Hello, I did some performance tests with uBLAS and MTL. Here are some results for those who is interested. The following tests were run: 1) uBLAS dense matrix multiplication: - row_major * row_major: ures.assign(prod(ur, ur)) - row_major * column_major: ures.assign(prod(ur, uc)) - column_major * row_major: ures.assign(prod(uc, ur)) 2) MTL dense matrix multiplication: - row_major * row_major: mult(mr, mr, mres) - row_major * column_major: mult(mr, mc, mres) - column_major * row_major: mult(mc, mr, mres) 3) C matrix multiplication (basic multiplication algorithm for C array). 4) uBLAS sparse matrix multiplication (20% of non-zera elements): - row_major * row_major: ures.assign(prod(ur, ur)) - row_major * column_major: ures.assign(prod(ur, uc)) - column_major * row_major: ures.assign(prod(uc, ur)) ures - is dense matrix. 5)MTL sparse marix multiplication (20% of non-zera elements): - row_major * row_major: mult(mr, mr, mres) - row_major * column_major: mult(mr, mc, mres) - column_major * row_major: mult(mc, mr, mres) mres - is dense matrix. All tests were run on Windows2000, I used gcc 3.2 with -O3 optimization flag. boost 1.29 was used. Here are some results: 1) uBLAS and MTL have approximately the same performance for dense matrix mult. uBLAS is a bit faster with small matrices (< 50-100), MTL is faster with large ones (>100). 2) When working with C array multiplication it is 5-6 times faster than both uBLAS and MTL. (!!!) 3) If I use my own simple mult funcion to mutiply uBLAS or MTL matrices: template <typename Mat> void mat_mat_mult(const Mat& m1, const Mat& m2, Mat& res, int rank) { for(int i = 0; i < rank; ++i) for(int j = 0; j < rank; ++j) { res(i, j) = 0; for(int k = 0; k < rank; ++k) res(i, j) += m1(i, k) * m2(k, j); } } It works 2 times faster than native uBLAS and MTL implementations of multiplication. (Iterators overhead?) 4) Dense matrix performance doesn't depend on row orientation neither for uBLAS nor for MTL. 5) MTL sparse matrix mult doesn't depend on row orientation and in any case much (2 times) faster than the best case for uBLAS. 6) uBLAS gives best performance for sparse matrices in row_major * column_major case (it is 3 times faster than column_major * row_major). 7) When I tryed latest uBLAS from boost cvs. It worked much (up to 2 times) slower for sparse matrices than 1.29. I would need to do other tests to make any real conclusions, but preliminary it seams to me that abstraction penalty for both uBLAS and MTL is too big. Alexei.

4 5

boost/jam/solaris
by gpolicello 22 Nov '02

22 Nov '02

Hello, I'm trying to port a simple app written in MSVC for NT to Solaris. The compiler we are using is Sun Workshop. How do I build boost? I've downloaded bjam_solaris but it won't compile. Apparantly, it is bringing in gcc-tools. Am I going to have to write a new tools file for Sun's compiler? Is there a simple make file I can use instead of jam? Thanks, - George

1 0