[cairo] Clip invariance issue

[Søren Sandmann]

Søren Sandmann <sandmann at cs.au.dk> writes:

> A typical max difference is around 30, which seems quite bad even if it
> is decided that cairo will not guarantee pixel-exact matches for this
> usecase.

I collected some statistics on the max differences:

  difference 24 happened 45 times, for example in test 1907
  difference 25 happened 231 times, for example in test 1985
  difference 26 happened 330 times, for example in test 1998
  difference 27 happened 258 times, for example in test 1991
  difference 28 happened 192 times, for example in test 1970
  difference 29 happened 215 times, for example in test 1995
  difference 30 happened 82 times, for example in test 1962
  difference 31 happened 197 times, for example in test 1997
  difference 32 happened 182 times, for example in test 1999
  difference 33 happened 223 times, for example in test 1989
  difference 34 happened 18 times, for example in test 1993
  difference 35 happened 6 times, for example in test 1520
  difference 36 happened 1 times, for example in test 1250
  difference 37 happened 1 times, for example in test 746
  difference 42 happened 1 times, for example in test 1495
  difference 46 happened 1 times, for example in test 1904
  difference 58 happened 1 times, for example in test 1681
  difference 73 happened 1 times, for example in test 894
  difference 84 happened 1 times, for example in test 549
  difference 85 happened 1 times, for example in test 1757
  difference 102 happened 1 times, for example in test 172
  difference 119 happened 1 times, for example in test 1878
  difference 142 happened 1 times, for example in test 623
  difference 145 happened 1 times, for example in test 777
  difference 150 happened 1 times, for example in test 481
  difference 151 happened 1 times, for example in test 1818
  difference 175 happened 1 times, for example in test 72
  difference 254 happened 1 times, for example in test 1939
  difference 255 happened 5 times, for example in test 1936 


Søren