... I don't believe it's feasible. When implementing the bindings I found it to be very subtle. For instance, lapack functions might have different signatures...
Hi, After some discussions in my group when we had Jeremy Siek visit this weekend we agreed that we would like to see generic C++ lapack interfaces that work...
Hi Matthias, hi all, ... This is the main idea behind the vector and matrix traits classes in the bindings library -- for example, currently blas level 1...
Kresimir, Mathias, all, Already from the experience of implementing the current bindings I learned a lot of things through cooperation with Kresimir and I...
... That's a good idea - it also helped me in other projects to discuss interface issues with others and am happy to participate. ... I just checked it out but...
Hi Stephan, ... 2; ... The second slice 'slice (1, 0, m.size1()-2)' is degenerate in specifying a stride of 0, but a size of 2. With the modification //...
... in boost-sandbox/libs/numeric/bindings/traits/doc/index.html ? I doublechecked and it's there ?! ... yep , that's him. ... Where do we need types with...
Joerg, I've mentioned this before but never followed up the issue : matrix<T> has a clear() member function but the proxies do not. Should'nt all proxies...
... I found oth the index.html and the fortran.h file but did not see fortran.h mentioned in the documentation ... If we just want to bind to blas/lapack/atlas...
Salut, ... at least one could do something ugly like #define USE_LAPACK_UNDERSCORE_IN_NAME #define USE_LAPAPCK_CAPITAL #ifdef USE_LAPACK_UNDERSCORE_IN_NAME ...
Salut, ... sorry, I just realized that the Daxpy in the second last line does not get substituted. Maybe it's easier to write a small C++ programm that ...
Hi Toon, Hi Mathias, hi all, Toon wrote: [...] ... [...] ... [...] I must admit that Fortran is not my strong point, so I'll leave that part to you -- but, of...
Hi Joerg, hi Stephan, ... It also fails with como. As it should. This is the old problem of binding references to temporaries: parameters of `swap()' are...
... They are indeed orthogonal, but you need to know the requirements (which you probably due): pointer to first element, stride, leading dimension, size. ...
... This might indeed be the easiest but it is such a pity that we have to use C++ executables to create C++ code instead of using preprocessor hacks or TMP...
Hi, I'm not sure, because I never used it, but GNU AutoGen can convert characters: `string-downcase' - lower case a new string It creates a new SCM string...
Leonildo Tincani
leonildo.tincani@...
Apr 7, 2003 6:31 pm
611
Hi Matthias (hi all), ... I've been thinking about the lapack bindings for some time now and I believe it's firstly necessary to limit the scope of...
Hi Kresimir, ... Thanks for checking that! ... Not sure about that. Maybe swap()'s signatures are inadequate for the view classes. If I change matrix_row<>'s...
Hi Joerg, ... Is there not an efficiency issue with the additional copy constructors necessary for the pass by value? Or are you assuming (hopeing) the...
Salut, ... Does it make sense to have a call by value for swap ? Why should I want to swap local copies of matrix. I expect from a call by value that the...
... Compiles with VC++ 7, boost 1_30. VC7 allows temporary structs as non-const reference parameters, but no ints. I seem to remember a rumour that the ...
... matrix_row<> object contains only a reference to the matrix and a row index, so copy constructor is not very expensive. Instead of copying pointer (which...
Salut, ... With that semantic, one should use swap() wit copy semantics, since r1 and r2 are not changed. I thought that the references should be exchanged,...
Salut, ... actually there is a swap_ranges in STL which swaps parts of a container, see http://www.sgi.com/tech/stl/swap_ranges.html . Best regards, Peter...
... Obviously, there are two possible semantics of swap for proxy classes -- swapping references and swapping actual elements. Question is whether the exchange...
Hi Everyone While everyone is discussing bindings I thought I would look at one issue that came up way back in the Boost review :-) At the moment uBLAS uses ...
Hi Joerg, Hi Kresimir, ... and ... Since swap is usually assumed to be a fast basic operation even a small overhead associate with the copy constructor would...
Hi Michael, you wrote: [snip] ... We are talking about the intricacies of the language! I'm trying to get that stuff compiling on a compliant compiler. For the...
Having problems with message search?
Fill out this form to ensure your group is one of the first to be migrated to the new message search system.
