On Friday 12 September 2003 10:15, Jason Wood wrote: > The files were captured on a windows machine, but have not been edited, so > they will be pure raw DV files.
probably not, since that would not be conform to AVI specification and not be playable by MS Mediaplayer. But it's a detail, not so important for now. > Some of the files did not load in piave - on closer inspection, they are > the files that have been rendered by Premier. If you can provide me with a sample, I can try and take a look. > Using Kdenlive ;-) > > Sorry, it's difficult to me more specific - basically I was adding clips to > the project (this does not appear to be the issue), and then dragging the > to the timeline and manipulating them, playing them back, etc. My suspicion > is that the leak is within scenelist generation - on my home machine I have > "fixed" (with a hack for the short term) the issue with too many scenelists > being sent to piave, and the issue seems to have become less of one. I see. Well, to be true, I am amazed by how well things were working anyway ;-) I want to do the memory management right in the first place. Hence, I paid much attention to the mem.management of frames and buffers, but the render tree and more so the management of properties is not yet organized in a elegant way. So there are very likely to be many leaks. > > I think it would be much better if one could just dum the output to a > > file, I find the small list pretty unreadable. I don't use it anyway, I > > just start piave in a seperate shell. > > Hmmm, well there is an option to dump to a file, but no option to stop it > appearing on screen as well. I think I might take a look at the logging > capabilities on the Qt text widget again. The only trouble with that is > that they only appeared in Qt 3.1 - so it breaks compatability with KDE > 3.0, which is not something I want to do at the moment. Thet leave it. It's not important for me at the moment. > p.s. I find KJots (part of KDE somewhere) an incredibly useful application > for jotting down ideas - give it a try, it saves hunting for scraps of > paper :-) I'll check it out -- contacts: www.physi.uni-heidelberg.de/~dubitzky
