freenode/#clim - IRC Chatlog
Search
11:18:11
jackdaniel
(draw-text* pane (format nil "~%foo") 10 10) ; leads to out-of-bounds array access
11:20:47
scymtym
i agree. writing tests for eclector (and especially aiming for 100 % coverage) has revealed a ton of bugs - a few even in the implementations running the tests
11:23:00
jackdaniel
scymtym: there is a meme fitting what I wanted to say: "Ah, I see you're a man of culture as well" (https://www.dailydot.com/unclick/man-of-culture-meme/ :-)
11:24:29
jackdaniel
as of testing, a curious fact I've heard from somebody is that statistical research shows, that usually most interesting bugs (in quantity and impact in published software) come from connection of two orthogonal modules (not in one module nor in more than 2 modules; some bugs in such combinations occure too of course)
11:25:13
scymtym
idiomatic use seems to refer to more "depraved" (the article's word) activities than testing and finding bugs, though
11:25:55
jackdaniel
well, this link I've provided is wrong, my intention and understanding of this meme is different (i.e w/o irony etc), but the image is correct
11:26:24
jackdaniel
I just wanted to link to the picture, but noticed explanation and assumed, it will match my expectations, which it did not :(
11:27:57
scymtym
regarding interesting bugs. in my experience, even simple unit tests reveal interesting bugs if ones goes for full coverage. case in point, .. is illegal, .||. is a symbol
11:52:44
jackdaniel
and the second one has :scroll-bars nil and is wrapped in scrolling in layouts section
11:53:09
jackdaniel
in the first application pane when I scroll design ink remains stuck to a position relative to a viewport
11:54:00
jackdaniel
(screenshot doesn't show this, but both are scrolled by 10px by clicking a scrollbar button)
14:39:07
jackdaniel
here's a sketch of word-wrap for multiline draw-text* which works on both seos and raw mediums: http://hellsgate.pl/files/draw-string.txt (it trims outer spaces)
14:39:41
jackdaniel
when I'm done with patterns I'll think whenever we can integrate it with current codebase
14:51:13
jackdaniel
if it were depending on pane-size, then that would depend: if pane is not an output-recording-stream, then it would be repainted, so it would readjust itself
14:51:34
jackdaniel
otherwise, if it is an output-recording-stream that would depend on whenever pane is redisplayed on resize or not
14:52:37
jackdaniel
I'm testing on both basic-panes and extended-output-streams, and that is a good idea, because I was able to spot a few more bugs
14:53:02
jackdaniel
because one depends on raw draw functions while the other on replaying output records
15:20:25
jackdaniel
empty space on the right pane in the middle is due to use of the with-room-for-graphics with reversed y-axis, which doesn't work on master at all, and works badly on my branch (but I plan to make it work reasonably later)