On Wed, Oct 22, 2008 at 9:07 AM, Mike Marchywka
I had a couple of questions related to recent posts here.
Do you mean my posts or some recent posts on this mailing list in general?
The only part of boost I've used is the regular expressions and these, along with greta, work fine. I eventually did write some of my own code to handle special subsets when I had comparaitvely long vectors of patterns and targets. I'm not sure what you are looking for with either spatial index or FFTW but both of these bring to mind the idea of strategy selection that includes things like cache awareness. In the case of my regex vector problem, I guess had there been flags to set a strategy, even if one could not have been deduced with reasonable overhead, that would help overall performance. I'm not sure that you want a library based on good progamming pracitce to be littered with implementations for quirky parameter ranges but the performance benefits can be worthwhile in some cases. If you are searching though a high-dimensional space, something like a DB query, it may be important to pick a strategy first and curious to know what there may be in boost related to this. Is there any kind of strategy selection or factory infrastructure? Even something as simple as a cache library, maybe similar to a thread library?
I don't quite understand what you are asking and how it is related with my original post. I asked spatialindex and FFT. Because I need to refactor some of my code written long time back, when there were no support of FFT and spatialindex in boost. I programmed by own spatialindex library (specifically, an R-Tree). I also wrote a C++ wrapper for FFT, which is far from perfect. If boost has R-Tree and FFT, I would use them rather than my own ones. BTW, would you please have an empty line between two consecutive paragraphs to make the post more readable? Thanks, Peng