the top of the map) works well. Depth issues have been taken care of as well
(sort of: same problems as non-clipped skybox, but you don't get sky IN the
map any more)
faces with 1 vertex are implemented, but I have no way of testing them :(ie, I
haven't found anywhere to test them, there just don't seem to be any sky polys
of the right shape)
be a bug in my face visiting code that's causing process_corners to think some
numver of faces < 5 are being visted when really 5 are. I'll sort that out
later (at worst causes a little overdraw).
cube vertex, that vertex is no longer added immediatly, instead delaying it's
addition to the poly vertex list until all the sky poly points have been
handled (this has a side benefit of completely eliminating those two almost
identical functions: enter_face() and leave_face() :). After the sky poly
vertexen have been processed, the faces the edges of the poly have visted are
analysed for all 14 possible cases to determine how many and which cube
vertexen are to be added to the face polys. So far, 9/14 cases are handled
properly and 1 seems to be mysteriously handled in what seems to be a correct
fashion part of the time (bloody heisenbugs). The remaining cases are: 5
visited faces with 1 vertex and 3 vertexen (the heisenbug); 6 faces and 2
vertexen (2 variants: neighboring vertexen and opposing vertexen); and 7 faces
with 1 vertex.
interface. To detect an extension, call QFGL_ExtensionPresent(). To get
the address of an extension function, call QFGL_ExtensionAddress(). This
function uses the preferred method for finding a function, based on
whatever OS and windowing system you use.
changed all parms references to directly access host_parms to improve readability, and removed unused basedir references. (all remaining fields in quakeparms_t deserve to be there)
player connects immediately after a timekick sample, their time WILL be
off, so don't look at them the first sample.
sv_user.c: Another location to enforce sv_maxrate, make sv_timekick and
friends ignore a user's time if it's -1, and apply double fuzz to times
lower than we expect -- it's way too sensitive otherwise. Also, some
whitespace changes.