00:00:00 --- log: started retro/12.01.03
00:48:19 <ok2> I have also written an full featured Common Lisp implementation of Ngaro VM, it is not fully done yet, and not fully working or tested, but i working on it, here is source code: http://tinyurl.com/7bmep27
00:50:30 <ok2> it should mostly work on SBCL :-)
00:54:24 <ok2> oh, it works on CLisp to, but not on ECL ...
07:24:45 --- join: karswell (~coat@93-97-29-243.zone5.bethere.co.uk) joined #retro
09:03:50 <crc> ok2: looks pretty good. it passes the core tests except for the divmod (#19) instruction
10:04:48 <ok2> crc: which one? :-)
10:46:18 --- join: dizm (~dave@121.98.168.127) joined #retro
11:02:41 <crc> specifically, division involving negative numbers
11:03:31 <crc> try:  -5 2 /mod .s
11:03:48 <crc> the results should be:  <2> -1 -2
11:04:06 <crc> with your lisp implementation, I get:  <2> 1 -3
11:21:27 <ok2> ok, fixed now, common lisp have 4 operations which do the same as /mod operation in retro forth, and they handle the positive/negative numbers different, i think i have found now the right one :-)
11:22:28 <ok2> the next thing is to test and fix the file operations, but not today ...
12:08:25 <ok2> i have also commited now some optimizations, it is now not far away from the C version in performance, at least on calculating fibonacci numbers
12:09:32 <ok2> i have tried this:  : fib ( n-m ) dup [ 0 = ] [ 1 = ] bi or if; [ 1- fib ] sip [ 2 - fib ] do + ; 32 fib
12:12:08 <ok2> my common lisp implementation needs for it 13.8 seconds, the normal retro VM in C needs 17.6 seconds, the previous common lisp needs 28.8 seconds
12:13:02 <ok2> retro VM in C with -O3 flag needs 7 seconds
12:13:23 <ok2> so i think 13.8 seconds is a good value :-)
12:23:51 <C-Keen> nice
13:16:45 <crc> ok2: revision 19 exits silently for me
13:17:09 --- nick: ivan -> Vladimir_K
13:37:16 --- quit: karswell (Ping timeout: 252 seconds)
13:37:46 <ok2> crc: oh, yes, i forgot, it does not stats atomatically, you need to call: (retro:ngaro)
13:39:23 <ok2> you can also do: (compile-file "retro-full") before (load "retro-full"), it will be much faster then
13:42:39 --- nick: Vladimir_K -> ivan
14:14:47 <crc> OK
14:14:59 <crc> I'll try it again
15:04:36 <retro_commits> new common lisp implementation from ok2 : http://rx-core.org/rx.fossil/info/d9a576f144cf02d4a11942ff813c111cf7af5746
15:08:45 <retro_commits> sync against latest arduino code from ok2 : http://rx-core.org/rx.fossil/info/ba0769a5a704ab2ed2a99686cf96507f0cf9123e
17:50:46 --- join: karswell (~coat@93-97-29-243.zone5.bethere.co.uk) joined #retro
21:33:38 --- quit: karswell (Ping timeout: 240 seconds)
23:59:59 --- log: ended retro/12.01.03