Gentlemen,
Having downloaded the nightly tarball for the first time in two
weeks and compared it to my private copy, I find several items of interest.
(1) Andy Ross submitted some font rendering changes ("PUI Rendering Fixes,"
sent on 5/4/04 at 2:15 PM) which didn't get into CVS.
(2) I've mentioned the JS library once already this morning.
(3) Jonathan Wheare submitted some changes to PSL ("psl copy constructor,"
4/20/04, 9:03 AM) that implement reference counting. Are we planning to put
those into CVS?
(4) My copies of "psl.h" and "pslProgram.cxx" include a variable "is_copy"
in the "pslProgram" class. Was this overcome by item (3)?
(5) I have code in "puAux.h" and "puAuxChooser.cxx: that implements a
"puaChooser" widget. Steve created this widget for his Triereis simulation.
If somebody could please put it into CVS for me that would be a good thing.
(6) The code in PW for making the Windows key auto-repeat act like the X key
auto-repeat hasn't made it into CVS either.
(7) I was trying to get multisampling to work in "pwWindows.cxx" but didn't
get anywhere. Any help would be appreciated.
(8) There is a minor efficiency change in "sg.cxx" and "sgd.cxx" on lines
863-865 and 882. First, the definition of "l" and the call to
"sgLineSegment3ToLine3" are unnecessary. Second, rather than call
"sgNormalizeVec3" on line 865 (which involves a square root), it is more
efficient to divide by "sgScalarProductVec3 ( v, v )" in the "return"
statement on line 882. The magnitude of "v" on lines 868 and 875 does not
matter because all we are checking is the sign of the dot product.
(9) I have a few efficiencies for "sgIsect.cxx" and "sgdIsect.cxx" which I
have gone over before and have probably beaten to death.
(10) The file "ssgVtable.cxx" contains code starting around line 562 to tell
whether a triangle intersects a sphere. I have a more efficient version of
this if anybody is interested. I also wonder whether the capability belongs
in "sg" as it is perhaps general.
John F. Fay
joh...@eg...
|
