11 Nov
2005
11 Nov
'05
6:11 p.m.
Quoting Maxim Shemanarev <mcseem@...1076...>:
I don't see any other elegant way of detecting the loops, because they can have any level of indirection (a pattern refers to another pattern that refers to another pattern, etc... and the first pattern refers to the first one).
See src/sp-gradient.cpp:chase_hrefs for one approach.
(At the time I wrote that, I wasn't aware of §5.3's prohibition on all circular references: unlike circular references for <use> or the <pattern> example that mental gave, circular references among gradients would seem reasonable but for §5.3's prohibition.)
pjrm.