01:48 FROGGS_ joined 01:59 FROGGS[mobile] joined 02:06 btyler joined 04:20 woosley left 06:18 woolfy joined 06:49 FROGGS joined
tadzik timotimo++ 07:01
that bit me with steroids :) 07:02
jnthn Bit me when making demo app with Gtk::Simple too ;)
But I could avoid it in such a simple thing :)
Glad it's fixed.
07:05 zakharyas joined 07:42 brrt joined
brrt jnthn, with a (small) bit of asm, making a stack frame out of heap memory is actually really trivial 07:55
(backlogging from the discussion yesterday) 07:56
case in fact
you could use dynasm to compile a 'move to heap, call and return' wrapper
if you were so inclined 07:57
(yes i'm aware of the evil, but calling compiled functions sensibly is a problem i'll have to solve /anyway/)
jnthn brrt: Yeah, but then we need to do it on all places we want to support with no sane non-assembly fallback :) 07:58
brrt all the world is x64 right :-p 08:02
i know
unsolved problem
by the way, i think i've learned a new heuristic this week, or at least how to better formulate that heuristic 08:05
'don't do by algorithm what can be done by data'
jnthn Hmm 08:07
Example? :)
brrt its an R example, if you care for it :-) 08:08
basically, i had a data frame (=~ a 'table') containing two columns of interest : 'type' and 'condition' 08:09
and i needed to calculate some values with respect to both these columns
now for just those two, thats not very hard, R will give you a matrix (= also atable) 08:10
however, i needed to relate them to a time variable as well - i.e. calculate both over time
which gives you a three-dimensional matrix 08:11
and even that isn't much of a problem but that many columns of our three-dimensional matrix are empty completely (because they did not appear in the actual experiment)
and i was very busy writing the logic for simplifying this 3-dimensional matrix to two-dimensional form 08:12
when it struck me that i could just paste the relevant columns of type and condition together
i.e. i replaced an algorithm with a data operation
jnthn :) 08:13
Sounds like a nice simplification.
08:14 lizmat joined, zakharyas joined
brrt well, yes :-) 08:15
08:15 lizmat joined
brrt especially as R is not the most algorithmically elegant language imaginable 08:16
08:19 zakharyas joined, woolfy left
dalek href="https://moarvm.org:">moarvm.org: 746e1a7 | jnthn++ | / (4 files):
Various updates for 2014.05 release.
09:47
jnthn www.moarvm.org/ updated :) 09:50
FROGGS that reads very well :o) 09:52
jnthn brrt: I figure at some point soon we should do some more detailed planning for your JIT work. I figure so far you're working on understanding the ingredients involved. I guess you'll have some questions on spesh too. :) 09:53
It'd be good to figure out concrete tasks from here. 09:55
10:13 FROGGS[mobile] joined 12:39 vendethiel joined 13:36 vendethiel joined 14:04 btyler joined 16:00 FROGGS joined
timotimo jnthn, got an idea for what to prioritize in gtk-simple? 18:40
or maybe something for spesh or mvm in general? 18:41
not sure i will be able to finish the named parameter spesh thing on my own
jnthn timotimo: Well, adding more controls and events? :) 18:42
I know we're missing a bunch of popular ones (like, check boxes, radio buttons, sliders, progress bars, images...) 18:44
tadzik timotimo: porting the old niecza gtk# examples might be fun :) 18:47
also, I can check if I can port my bachelor's app to Perl 6 now
timotimo something that can draw stuff would be a nice base for turtle graphics
tadzik uit's also using gtk for gui, I can make a list of things that it needs
timotimo how about dbus?
tadzik hm, using gtk for the last 5 years I never touched dbus :o 18:48
timotimo does glib have something C ish for that?
tadzik there's glib-dbus orso
timotimo sounds helpful
it's good for system utilities
well, desktop utilities, really 18:53
at some point splitting the code into filesbwould be interesting
tadzik looks at his gtk programs to look for something easy to port 18:57
oh, github.com/tadzik/l should be simple enough :)
timotimo justba launcher? 19:05
FROGGS timotimo: what happened to your spacebar? 19:13
timotimo this is a phone keyboard 19:20
if i were at home or at least had my laptop with me, i'd be doing p6 stuff 19:32
tadzik yeah, a fast launcher :) 19:35
timotimo 'fast' ;) 19:38
tadzik it is superfast, 'cos it has no startup time 19:46
it runs in the background, and then SIGUSR1 makes it show up
timotimo the perl6 port will be super fast, too? 20:06
[Coke] github.com/tadzik/l - readme has a formattating error. your literal `'s got turned into {{MONOSPACE}} 20:09
*formatting
tadzik maybe :) 20:30
[Coke]: it did? I don't see ti
22:01 lizmat joined 22:04 woolfy joined 22:25 woolfy left