From b3c3167a492490c5872850c7df2cb56223003b52 Mon Sep 17 00:00:00 2001 From: =?utf8?q?Andr=C3=A9=20Wobst?= Date: Fri, 16 Apr 2004 08:03:55 +0000 Subject: [PATCH] some comments to ongoing debates git-svn-id: https://pyx.svn.sourceforge.net/svnroot/pyx/trunk/pyx@1678 069f4177-920e-0410-937b-c2a4a81bcd90 --- CHANGES | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/CHANGES b/CHANGES index d7d31f19..dcd0b729 100644 --- a/CHANGES +++ b/CHANGES @@ -38,16 +38,28 @@ TODO: - deco module: - think hard about different paths in decoratedpath class - adjust parameters of cycloid and smoothed - AW suggests to use cycloid with default curvesperloop=5. + AW: I suggest to use cycloid with default curvesperloop=5. - dvifile module: - support single-character mode - path module: - arclentoparam should raise an exception when the arclen is out of range - it seems to be, that epsilon is a feature of normsubpath; - Why is there (also) a self.epsilon? + AW: Why is there (also) a self.epsilon? JL: What class does self refer to? + AW: I'm sorry for not being precise. Both, normpath and normsubpath do + have an epsilon. I'm not sure if we should do that (its slightly + confusing). In normpath._findsubpath we should refer to the + epsilons of the subpath. The only usage of epsilon left in normpath + is in the constructor and in append (BTW -- we do have some code + dublication here). A possibility would be to add a keyword-argument + epsilon to append. The point is, that I would like to get rid of + self.epsilon, where self is a normpath instance ... - How about normpathel.begintrafo() and normpathel.endtrafo()? JL: -1 + AW: May I be able to change your rank to -0 when comming up with a + profile for typical use cases of (new style) boxes? We do have + begin, at and end for the normpathel's. Why is the analogous for + normpathel.trafo a bad idea? - color module: - what should we do for palette when only one entry -- 2.11.4.GIT