On Feb 22, 2013, at 11:23 AM, Louis Wasserman wrote: > Sweet. Just be aware that the floatValue() optimization requires > Float.parseFloat to be fixed to match IEEE 754 behavior, as you're apparently > doing with Double.parseDouble, to pass tests.
Actually I missed that until yesterday but found it out myself. I do not think however that the patch for 7131192 necessarily has to precede the fix for 7192954, as long as the latter gets into the same release. > (The separate patch I sent in to fix that is probably invalidated by your > changes, but it shouldn't be too difficult to do yourself -- it really is > just copy/pasting the Double.parseDouble implementation and using updated > constants.) Thanks for the pointer. As previously noted I am still trying to gain some traction in this area. Brian
