freenode/#clim - IRC Chatlog
Search
4:19:50
beach
This improved Bezier code could simplify other parts of McCLIM, especially for the framebuffer backend.
4:21:02
beach
If we have efficient code for rending bezier regions, then those can be used to approximate many others such as ellipses and circles. And it could simplify line drawings as well, because of things like caps and joins.
4:23:50
slyrus
hmm.. I'm assuming we can make true ellipses and circles fast enough that we don't need to approximate them, BICBW.
4:24:13
slyrus
but, yes, the lines and what not would probably benefit from being drawn as bezier curves
4:25:40
beach
We can make circles and ellipses fast. My point is that the approximation of such regions with Bezier regions are very accurate, and we could eliminate some code that then would no longer have to be maintained, tested, etc.
7:39:34
jackdaniel
regarding using bezier curves for approximating other things, we will have to mvoe bezier back to clim-internals in such case (if we want to do that), and reexport symbols from the extension for sake of clean bootstrap process
8:15:12
jackdaniel
hah, human subconcious is amazing! week after fixing bug in ECL I have realized that I have introduced by that yet another small issue
13:53:33
jackdaniel
clim spec doesn't provide a way to change list-pane-items, right? (I mean - we have that implemented in climi - shouldn't we provide it in extension or something?)
14:07:27
slyrus
jackdaniel: alexandria has ensure-list, but it does something different. I'll rename my function. Mine takes a sequence and returns it unchanged if it's a list, or makes a new list with the contents of the sequence
14:08:48
jackdaniel
OK, please rename it then (apparently I've got confused, small comment there what it does (and how it's different) will be nice - docstring maybe?)
14:20:27
jackdaniel
slyrus: will it be a problem if I'll assign my own pull requests for review to you? even brief peer review improves code quality and increases confidence that the change is OK
14:23:26
slyrus
so this +nowhere+ thing is interesting. something must have changed in the design/region initialization sequence. we reference these regions from design.lisp, even though region is a design and region.lisp gets loaded after design.lisp. what we were doing before was wrong, but I could just forward reference +everywhere+ (which was actually my first "fix" for this).
14:24:21
slyrus
I may have been too clever by half in trying to keep +everywhere+ as a constant-ish thing. It's a little disconcerting to lie to the users and just have everywhere be a normal (special) variable, not a constant, but it seems we do that anyway.
15:06:44
slyrus
no PR for you. now the only bona fide warnings are the %make-image-sheet-get-function thing I filed an Issue on and Drei, which I assume we won't fix.