set_at now take a long long for y instead of int #2962
Conversation
…ve the memory for it
oddbookworm
force-pushed
the
fix-draw-line-segfault
branch
from
c612961
to
1dc41d6
Compare
| def test_line_draw_large_surf_regression(self): | ||
| """Regression test for https://github.com/pygame-community/pygame-ce/issues/2961""" | ||
| try: | ||
| surface = pygame.Surface((14457, 37200)) |
There was a problem hiding this comment.
I don't like the addition of this test. It makes the assumption that the host has enough memory.
A better solution would be a RAM check with pygame.system.get_total_ram, and then running this test if the RAM is above some threshold, but I'm not a fan of this either.
There was a problem hiding this comment.
I don't particularly like it either, but I felt it would be better to have some way to test that it hasn't regressed in the future. I couldn't think of any solution other than "make a gigantic surface" that CI won't have the memory for. I did try to make the surface as small as I could and still cause the segfault without the changes from this branch
There was a problem hiding this comment.
A follow-up of my message, yes I think it's not the method to follow to add the test.
|
I don't like how the problem is solved, we aren't clamping the line in the clip rect so we have to check if each pixel of the line one by one are outside the clip rect. IMO, the best way to solve this problem is to delimit with new coordinates which part of the line is going to be drawn. |
Fixes #2961
@ankith26 figured out that it was an
intoverflow when multiplyingsurf->pitchbyy