From: Alan W. I. <ir...@be...> - 2002-09-23 18:16:28
|
I have recently run across an orientation bug for non-integral -ori values. If I try ./x12c -dev xwin -ori 0.2 and hold a piece of paper up to the screen it is immediately obvious that the box has turned into a parallelogram with corner angles which are noticeably different from 90 deg. This problem also occurs for the psc and png devices. If you would like an extreme case of this problem try a large aspect ratio such as ./x12c -dev xwin -ori 0.2 -a 10 If you want the problem to completely disappear try an aspect ratio of unity. ./x12c -dev xwin -ori 0.2 -a 1 I suspect the core is doing something dumb like applying a rotation transformation to normalized device coordinates (whose x,y ratio varies from true device coordinates by typically a factor of 3/4, but you can make this 10 or unity by the above manipulations of the aspect ratio) rather than actual device coordinates. However, there is no way I will ever understand the core rotation code so I cannot fix this myself. Maurice, could you have a look at this to see if there is a quick fix to this "parallelogram" bug? BTW, the reason I ran into this was I was comparing how the postscript fonts in the ps.c code and freetype.c (Andrew's version which he should shortly be checking in) handled non-integral -ori values. The results currently look ugly because both codes rotate the fonts properly (x and y labels written at 90 deg with respect to each other) while the core code does not rotate the plot properly (box angles not 90 deg unless the aspect ratio is unity). Alan email: ir...@be... phone: 250-727-2902 FAX: 250-721-7715 snail-mail: Dr. Alan W. Irwin Department of Physics and Astronomy, University of Victoria, P.O. Box 3055, Victoria, British Columbia, Canada, V8W 3P6 __________________________ Linux-powered astrophysics __________________________ |