freenode/#clim - IRC Chatlog
Search
19:37:54
jackdaniel
also, you may want to use with-drawing-options instead of setting and reverting clipping region manually
19:41:23
scymtym
that's good to know but using clipping didn't work because the output-record bounding region seem unaffected by clipping. that messes up the layout. this problem is the reason for only drawing function names that fit entirely into the rectangle
19:54:06
nyef
Why would the bounding rectangle of an output record have anything to do with a clipping region?
19:57:07
jackdaniel
nyef: if you have presentation which is active, we test against bounding rectangle not the clipping region, so it may get selected even if user goes to the area which is not drawn (as far as he can see)
19:58:23
jackdaniel
also, if we clip two rectangles which width is 4 to 2, you could fit two such rectangles in 4-wide area, but we lay them out according to bounding rectangle
20:00:29
jackdaniel
rectangle might have been not the best example, better one is the application scymtym provided
20:17:37
jackdaniel
but I can't see how that would be implemented to avoid these two problems at the moment
20:19:19
oleo
hmm if scaling is involed do nothing (other than signified by something else to not) if no scaling and rectangular then implement the same hit test for rectangles
20:38:08
jackdaniel
I've manged to make it clip bounding-rectangle as well: http://hellsgate.pl/files/b87819e3
20:39:25
jackdaniel
oleo: pasting these links isn't helpful, I'm sure I've read these passages at least a few times each in the past
20:40:00
oleo
hmm, i see a distinction between drawing on the medium of the sheet and directly to the sheet
20:42:52
oleo
hmm, so if nothing was supplied by the programmer the default is the medium provided one but on replay things could have changed
21:58:07
nyef
Why should stop-the-world GC require effort-to-stop proportional to the number of threads, rather than proportional to the number of CPU cores (thread execution engines)?
22:01:24
nyef
I know that it typically takes effort proportional to number of threads. But *why should it*?
22:01:42
oleo
don't you have to impend the effort on all threads rather than just some from the chores
22:05:13
nyef
Any thread not actively running is already stopped. Why go to the trouble of waking it up in order to put it back to sleep?
22:06:49
nyef
Or a thread that has exhausted its timeslice and the scheduler hasn't put it back on a CPU yet. It doesn't need to be stopped, it just needs to be prevented from restarting.
22:08:56
oleo
that would clog the pipeline of the scheduler, and you'd need a watchdog for each of'em
22:10:04
oleo
to prevent from restarting you have to always steal cpu from it, like in you have to watch over it
22:11:44
oleo
for one it would not be a problem but for many you'd at least need another thread/watcher....
22:13:08
oleo
and being external and having to count which ones you have to count over and bookkeeping, there is the original proces which spawned it all and has the history....you don't need that
4:21:18
beach
jackdaniel: I think I promised something I can't keep. I have lunch guests today, and I am going to be busy cooking pretty much all morning.