Just for practice, I decided to go oldschool-logo and do up a Sierpinsky Triangle app with the pen controls. It's mostly working, but... the top of the first iteration and only the first iteration is offset left by about 6 pixels. Odd.
Adding keyed waits and paying close attention to the variables, it appears that the midpoint of the first horizontal line is being calculated at -5.68e-14. As soon as that variable is used to set the sprite position, however, it becomes -5.7, or -6 when rounding to the nearest pixel, as if the exponent is being completely dropped.
I just uploaded the example to the server, it may even come up preset to the problem spot.
...
This might also have to do with storing stuff in lists, if (as came up in the lists of lists thread) list contents are literalized and then parsed back. In that case, if the parser can't handle e, rounding negative exponentials to 0 and positives to MAX_INT upon literalization would probably be a preferred behavior.
Last edited by PsychScratchpad (2009-03-21 19:49:34)
Offline