freenode/#clim - IRC Chatlog
Search
6:04:33
beach
I accidentally entered some comments in the #clasp channel. Can you look there, or should I retype them here?
6:09:53
ck_
I don't think there should be only a triangular hole with :NONE as joints, because the segments would be connected at their endpoints -- which is the middle of the stroke, right? not the corner
6:17:04
ck_
you're right. I was somehow picturing different widths for the two segments, with an overhang at the other side
6:20:09
beach
I am thinking two things: 1. Since the CLIM specification says that this stuff is optional, we should feel free to do it right. 2. For each case (line style, polyline, polygon, filled, not filled) we should decide on a shape defined by paths (possibly Bezier splines) that defines the "meaning" of that case.
6:20:17
ck_
loke: if I whip up a naive triangulator for those cl-vector polygons, can you try it out sometime? Just for fun
6:23:49
beach
Like in the case of :NONE, is the overlap drawn less translucent if the ink is translucent?
6:25:20
ck_
from the perspective of a user, let's say an artist, who wants to put some art on the screen: I don't think so
6:26:40
ck_
but no, I don't think so. Maybe there is some more ink on the parts of a stroke with a fountain pen where you stay longer, but that _definitely_ shouldn't be in the shape of parts of overlapping polygonal geometry
6:27:53
ck_
so, if you want to draw overlapping parts, use separate strokes -- that produces separate triangulations which will then blend inside the rendering framework
6:37:07
beach
I am not going to argue with you or agree with you, because I don't have a preference. My point is just that all this has to be decided.
6:41:24
beach
Though, I maintain that, for each case, we should decide what shape(s) the case corresponds to. Then we can generate that shape, triangulate it, and then render it.
6:49:02
beach
I mean, if we take loke's example, there should be semicircle -- straight line -- whatever join curve -- straight line -- semicircle -- straight line -- straight line -- back to start.
6:54:23
beach
I have no strong preference, other than that it is probably best to make a decision so that everyone knows what is being worked on.
6:54:31
ck_
like I said, I'll write some demo code for that when I get around to it, then we can decide
6:58:52
ck_
Not that it's required for a decision of course, don't get me wrong ;) but right now I have to go do something else.
7:36:37
loke
ck_: yes of course. I have the overall infrastructure implemented already, and it should be a simple thing to plug it in
7:43:05
ck_
loke: okay I'll get back to you with that then. Probably your tomorrow though, at the earliest
12:45:15
jackdaniel
(branch in which I'll try to make backends work without any notion of frame, requires only few changes)
12:46:13
jackdaniel
as far as I can see in various "Liberation Front" names, the subject being liberated goes first
12:54:28
ck_
I think I did everything you asked (except for the case/cond thing) in #792 but one of the commit comments does not show as outdated even though it is -- I don't understand
12:54:52
ck_
maybe github is too advanced for me and I should go back to mailing lists and patch files
13:01:49
jackdaniel
yes, I was about to say that (but I've decided on the liberation front joke), that I have a feeling that we are gaining mass
13:54:54
ck_
I had hoped for something a little over the top, you know, with the python humor still reverberating in the background
13:55:29
ck_
Yes people -- with progress like this, I think that sometime in the year two-thousand... we'll be able to totally reduce the list of issues to maybe one or two .. pages
13:55:55
ck_
especially with commits like the other day, from that guy, with all the parentheses in it.. that was awesome
13:58:17
jackdaniel
over the top you say? with our speedup we'll reach the level of technology from 1988 in about 5y \o/
17:09:53
anaphoric
I'm curious about graph layout/drawing algorithms - especially the one in CLIM (like this example https://upload.wikimedia.org/wikipedia/commons/c/c6/Listener.png ). Does that algorithm have a name?
17:12:25
ck_
looks a little bit like the one knuth described -- if you look at the leaves, you can count one line per.. then go back from there