Hi Michael, ... Thanks that you still take the time to contribute so much! [snip] ... mutable ... We've got three element proxy classes: sparse_storage_element...
Hi Michael, ... I'm rather unsure about this. My only suspicion is that the historical formats could have a bad locality of reference when accessing the...
Hi Joerg, ... Standard containers also have different requirements regarding invalidation -- it can be said that std::vector<> is very unsafe ;o) Obviously, it...
Hi again Kresimir, ... Yep this is very much at the root of the problem. I think I am a bit fixated on std::map because it is wonderfully safe. No invalidation...
Hi Joerg, ... I just couldn't stop myself. I think it will catch on. It has with me! ... That is where uBLAS comes it. ... Sorry Joerg, while I was writing...
Hi Joerg, ... Not very, on second thought ;o) Let's look at the code: =========================================== void append (size_type i, const_reference t)...
Hi Michael (hi, all), ... assignment ... [snip] ... The first series of CLAPACK tests is done. Due to the reduced fill-in I've found and (hopefully) fixed a...
Hi, I've updated GluCat to use the uBLAS version from the current Boost. I've tested it with gcc 3.1 (SuSE pre-release) and Intel C++ 6.0.1. The Gzipped Tar...
Hi Joerg, I've had a quick look at, and regression tested uBLAS_11_02. Wow look like you did a lot! Testing with Bayes++ shows your include fixes for Sparse...
Hi Michael, ... data.size() may be 0, too (as I always relearn testing with CLAPACK ;-). ... Is this an important QoI issue currently? ... Yep, I noticed this...
Hi Joerg, there is a bug in the BOOST_UBLAS_APPEND_ONLY branch of insert(). Test in `if' (to insert or not) on line 2629 is: if (filled1_ == element1 + 1 &&...
Hi Kresimir, ... Yes, I've to confirm this bug ;-( ... How do you like something along the lines of (filled1_ == element1 + 1 && (filled2_ == 0 || filled2_ ==...
Hi Paul, ... Here's my experience report: Installing/basic testing - Downloaded your port and un-zipped it. - Needed to have something like CPPFLAGS (=...
Joerg just a quick comment, ... I thought of this! If data.size() is zero then the only valid slice is slice.size() = 0. This is correct. When data.size() >0...
Hi Joerg, ... `filled2_ == 0' is IMHO not needed. Before the `if' is while (filled1_ < element1 + 1) { index1_data () [filled1_] = filled2_ + 1; ++ filled1_; }...
Hi Joerg, ... [...] ... [...] ... As I said in another post, this proxy stuff really became frightening and monstruous. Is it really worth all the trouble and...
Hi Joerg, ... [...] ... I'll include patches (I hope correct ;o) in modifications for variable index base. I hope that you will not work on compressed_matrix<>...
Joerg, Thanks for looking at this. 1 Where do I download the latest uBLAS? 2 What changes did you make to basis_element()? Thoughts below. ... and when ... The...
Hi Kresimir, you wrote: [snip index base related stuff] ... This probably isn't a bug, it works for me, if the preprocessor symbol BOOST_UBLAS_SAFE is defined...
Hi Paul, ... I usually upload it to the group's file section: http://groups.yahoo.com/group/ublas-dev/files/ublas_2002_11_04.zip ... Sorry for that missing...
Joerg, Thanks. I will download uBLAS and try a few things over the next few days. Also, my explanation of compressed vs full matrices in GluCat was not ...
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.
