"m...@apollinemike.com" <m...@apollinemike.com> writes: > Here's what does: > ----) Moving things to C++ > > There's still a lot of scheme code in stencil-integrals.cc (scm_foo, > scm_bar, etc.).
scm_xxx is not slow per se (the interpreter is more a cause for Scheme slowness than the primitives). However, Scheme data structures like alists are often not really access-friendly. > I don't know if this could be the reason for the slowdown. The > problem is that it's hard to get around this, as all of the data is > stored in Scheme structures. Depends on the kind of structure. For one thing, it appears to me that transform matrices should be smobified so that they don't get packed and unpacked all the time but are basically transparently passed around in Scheme and appear as a native type. For most operations on them, you should have a ly:xxx procedure that does not even bother unpacking individual elements. -- David Kastrup _______________________________________________ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
