Hi all,
1) Optimalisation: In gem-0.87, on linux, the makefiles in src subdirectories redefines the CXXFLAGS variable, instead of adding flags to it. Also there seems to be no optimalization flags set at all, unless this is - fPIC, but I can't find PIC defined anywhere. So I had to add these flags by hand to every makefile, not just to the toplevel one or as an environment variable.
2)Some of the geo's (f.i. polygon) have a maximum of 10 points. This limit is hardcoded. Is this an openGL limit or a GEM limit. Changing the GEM code to a larger maximum isn't so difficult, but the nicer solution would be to allow dynamic adding of vertices. Would that be theoretically possible? (it would probably be a new object)
Gerard
2)Some of the geo's (f.i. polygon) have a maximum of 10 points. This limit is hardcoded. Is this an openGL limit or a GEM limit. Changing the GEM code to a larger maximum isn't so difficult, but the nicer solution would be to allow dynamic adding of vertices. Would that be theoretically possible? (it would probably be a new object)
That seems like a reasonable change. You can dynamically add vertices with the current polygon object, it just has the 10 vertex limit. The problem would more likely arise from representing and handling large, dynamic data sets in pd without using something like an array. Maybe a vertex array object that uses a pd array for data is an option.
cgc
Gerard
PD-dev mailing list PD-dev@iem.kug.ac.at http://iem.kug.ac.at/cgi-bin/mailman/listinfo/pd-dev
chris clepper wrote:
problem would more likely arise from representing and handling large, dynamic data sets in pd without using something like an array. Maybe a vertex array object that uses a pd array for data is an option.
yes, but i think that pd's arrays are not ideal for this, because of low-dimensionality. i really do not like the way [linear_path] et al handle this. It doesn't allow you to interact with the data (like drawing in an array) Maybe we should extend pd's capabilities to arrays before ? (i'm not sure, whether the [matrix]-approach of zexy is much better: you cannot interact with the data, but at least it is ordered somehow meaningful. but it would be nice to have a real multi-dimensional structure for numbers)
mfg.as.dr IOhannes
vanDongen-Gilcher wrote:
Hi all,
- Optimalisation:
In gem-0.87, on linux, the makefiles in src subdirectories redefines the CXXFLAGS variable, instead of adding flags to it. Also there seems to be no optimalization flags set at all, unless this is - fPIC, but I can't find PIC defined anywhere. So I had to add these flags by hand to every makefile, not just to the toplevel one or as an environment variable.
i have changed this in the CVS now. CXXFLAGS will not be used directly, instead it is added to a $(cxxflags) variable. So you can use better tailor your make
Anyhow, i don't know why there are now optimizations i have added "-O3 -funroll-loops -ffast-math" for now. Maybe this will break something, then we can turn it off again
mfg.jzu.caw IOhannes
-
chris clepper -
IOhannes zmoelnig -
vanDongen-Gilcher