Well, using cachegrind now on RG takes forever - for example starting up RG with cachegrind and loading a file used to take about fifteen minutes to complete the trace - now it's about fifty minutes. I imagine this is directly related to how much extra calculation we're doing.
The cachegrind results are complex because there's so much going on of course. The big areas where we're spending time is in the calculation of chord rulers (as we know) and also the notation quantization when importing MIDI. Loading files generally and the subsequent calculations we're performing on them just seems excessive generally - I wait about five seconds for a file to complete loading - you try out any proper sequencer or notation editor (well, what are the stats here?) and the story is a little different. Sequencer users will expect instantaneous loading, I don't know what notation users expect - perhaps there's some trade-off we can make to improve this behaviour? cachegrind logs available if you want them - you only need to build kcachegrind to interpret them. R ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Rosegarden-devel mailing list [EMAIL PROTECTED] - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-devel
