On May 12, 2014, at 3:59 PM, Tom Browder wrote: > I am interested in that task and have searched extensively for the > original authors of that code to no avail so far:
I last wrote to Bill about ten years ago, but no longer have his contact information ever since he stopped running GICI. > In initially analyzing the file I found a likely error and commented > on it in the code at revision 60397 begging at line 845: > > It looks to me if there is a line of four values missing. I believe > the missing four values will result in the last four values being > initialized to zero, but that seems not to fit the rest of the data, > thus my theory of a missing line. Quite a find in the middle of that big table! I agree that this is a bug. The data will usually be zero for a debug build, but not for optimized builds (and there's no guarantee one way or the other per C). > The turb_table is used in one file (./src/liboptical/sh_wood.c) where > it is used to randomize tree ring dimensions. > > I am confident I can get a similar sub-sequence of a longer stream but > maybe not an identical one. Even if not identical I'm pretty sure > some output sample targets with the two sequences would meet the > looks-good-enough-alike test. > > Would such an almost-the-same sequence be acceptable? Given it's only used by the wood shader, I'd say yes. When I wrote the TODO, I'd incorrectly assumed that table was in use by more shaders than that. Frankly, the fixed table can probably go away completely to just call a random number routine. It'll be slower, but the wood shader It's "possible" that the table was generated with bn_unif_double_fill() or BN_RANDHALF(). If you do regenerate or swap the table with a function call, the shaders regression test will likely need to be updated. You'd have to make sure you're on a platform with no off-by-one differences first compared to the reference image. In all, fair game for change. Cheers! Sean ------------------------------------------------------------------------------ "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE Instantly run your Selenium tests across 300+ browser/OS combos. Get unparalleled scalability from the best Selenium testing platform available Simple to use. Nothing to install. Get started now for free." http://p.sf.net/sfu/SauceLabs _______________________________________________ BRL-CAD Developer mailing list brlcad-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/brlcad-devel
