00:00:00 --- log: started retro/06.08.26
00:02:32 <Quartus> Ok, I've retracted KEY, so I also had to retract ACCEPT and the 'ok' prompt. :/
00:03:29 <Quartus> Corrected version at http://quartus.net/retro/retro-ans.fs
00:16:40 <Quartus> I can't figure it out.  If I revector key to discard either 13 or 10, the stack depth is somehow screwed with inside the interpreter.
00:23:51 <Quartus> Ok, I've fixed it. I don't know why it happened, but I've moved the stack-underflow checking somewhere else.
00:24:16 <Quartus>  http://quartus.net/retro/retro-ans.fs
00:25:45 --- join: Cheery (n=Cheery@a81-197-19-23.elisa-laajakaista.fi) joined #retro
00:43:29 --- join: neceve (n=claudiu@unaffiliated/neceve) joined #retro
05:23:51 --- join: nighty (n=nighty@CPE00119576a9c5-CM0012c90d36fc.cpe.net.cable.rogers.com) joined #retro
05:25:45 --- quit: nighty (Client Quit)
08:33:54 --- part: crc left #retro
08:37:25 --- join: crc (n=crc@pool-70-110-129-33.phil.east.verizon.net) joined #retro
10:02:06 --- join: neceve_ (n=claudiu@unaffiliated/neceve) joined #retro
10:14:16 --- quit: neceve (Read error: 113 (No route to host))
10:18:15 --- quit: timlarson (Read error: 104 (Connection reset by peer))
10:32:55 --- join: timlarson (n=timlarso@user-12l320v.cable.mindspring.com) joined #retro
10:35:35 --- join: Quartus__ (n=Quartus_@209.167.5.1) joined #retro
11:02:16 --- nick: nanstm -> Raystm2
11:28:45 --- join: snoopy_1711 (i=snoopy_1@dslb-084-058-160-216.pools.arcor-ip.net) joined #retro
11:37:07 --- quit: Snoopy42 (Read error: 145 (Connection timed out))
11:37:21 --- nick: snoopy_1711 -> Snoopy42
11:45:56 --- quit: Quartus__ ("used jmIrc")
12:02:39 --- quit: neceve_ (Remote closed the connection)
13:00:40 --- quit: Cheery ("Download Gaim: http://gaim.sourceforge.net/")
13:35:13 <Raystm2> Chess is driving me nuts. 
13:35:28 <Quartus> chessnuts.
13:37:33 <Raystm2> yup. I broke it. The solution is not obvious. It's in the mechanics. I'll have to figure out how to break this down for testing.
13:37:45 <Quartus> Inline testing is your friend.
13:38:39 <Raystm2> Indeed. Time to insert some virtual breakpoints.
13:49:04 --- join: nighty (n=nighty@CPE00119576a9c5-CM0012c90d36fc.cpe.net.cable.rogers.com) joined #retro
15:45:04 <Quartus> With a small additional file of optimized stack words, the retroforth ANS layer now runs the fib benchmark about 20% faster than gforth-fast does.
15:48:26 <Quartus> Which is more in-line with what I'd expect of a native-code Forth, speedwise, at least one that doesn't do aggressive optimization.
15:49:17 <Quartus> The last 2x is always the hardest to get.
16:06:09 <Quartus> In this case it's the last 1.5x.  A hand-coded asm version of the bench runs 1.5x faster than the faster.fs+ANS layer.  The faster.fs speedups make a huge difference, without them the bench runs 2.6 times slower
16:56:00 --- mode: ChanServ set +o crc
16:58:51 <crc> good evening
17:05:41 <Quartus> Hi crc.
17:07:03 <Quartus> Ran a few numbers: http://retroforth.net/paste/?id=160
17:08:32 <crc> is faster.fs availible anywhere?
17:09:04 <Quartus> Yes, you're welcome to it -- you'll want to translate it into your own format, as right now it uses an ANS Forth assembler, unless you also want that.
17:09:56 <Quartus> If you do, let me email it to you, as I haven't cleaned it up enough to want to publish it.
17:12:00 <crc> yes, please send it
17:12:04 <Quartus> Ok.
17:12:38 --- join: ycombinator (n=wunderwa@bespin.org) joined #retro
17:13:02 <Quartus> Is there an email address you prefer?
17:13:28 <crc> crc@retroforth.org
17:14:26 --- part: ycombinator left #retro
17:15:12 <Quartus> Sent.
17:20:27 <crc> thanks
17:20:35 <crc> this should be useful :)
17:20:50 <Quartus> Sure. The assembler is ported from Gforth, just a small wrapper to add code/end-code really.
17:21:35 <Quartus> Nothing in faster.fs is set in stone except the new word definitions -- I was fooling around with si-- to see just how much I hated the notation.
17:23:08 <Quartus> I've ported the disassembler too, been using it to poke around inside rf.
17:23:51 <Quartus> The disassembler runs aground at the original-style embedded-value literals quite often.
17:26:32 <Quartus> still,
17:26:33 <Quartus> see .
17:26:35 <Quartus> ( $428BDD )  call    s>d  \ $E8 $E8 $F4 $FF $FF
17:26:35 <Quartus> ( $428BE2 )  call    d.  \ $E8 $E1 $FF $FF $FF
17:26:35 <Quartus> ( $428BE7 )  ret     near  \ $C3
17:26:38 <Quartus> Pretty neat.
17:29:43 --- join: snowrichard (n=richard@12.18.108.220) joined #retro
17:29:45 <snowrichard> hi
17:29:59 <crc> hi snowrichard
17:30:07 <snowrichard> hello
17:30:20 <snowrichard> what is the latest retro version? and how is it going? 
17:30:31 <crc> latest version is 9.2.9
17:30:43 <snowrichard> you have a native one ? or the linux?
17:30:59 <crc> 9.2 does not have a native build; 9.3 will though
17:31:10 <snowrichard> I'll grab linux I guess
17:32:05 <crc> development is going well; I am making some headway on cleanups to the core code, and expect to have a better overall language in 9.3
17:32:46 <snowrichard> got it I'll take a look soon
17:32:55 * crc is trying to keep up with the ANS code Quartus is writing for Retro
17:33:09 <snowrichard> maybe I should grab source too
17:33:19 <crc> source is always helpful :)
17:34:03 <snowrichard> ok I have a look at both
17:34:18 <Quartus> It's slowing down now crc, without >IN and SOURCE my hands are tied. :)
17:34:34 <snowrichard> I'm not familiar with the darcs tool
17:35:01 <crc> darcs is a version control app; it's easier to use than subversion or cvs
17:35:07 <snowrichard> really? 
17:35:08 <snowrichard> ok
17:37:16 --- quit: snowrichard ("Leaving")
17:38:42 <crc> Quartus: I am working on that :)
17:38:52 <Quartus> Good news. :)
17:39:06 <Quartus> That's take off the upper-limits on include file size, too.
17:39:24 <Quartus> And if a save-input/restore-input appears on the horizon, I can sew up the holes in eval too.
17:41:28 <Quartus> crc, one thing keeps happening -- sometimes, retroforth replaces the first character of the input line with an r.  So I'll type see foo and it'll say ree <- undefined word
17:41:43 <Quartus> This happens without the ANS layer, and randomly.  It's always an r, always the first char.
17:42:01 <crc> I have had that happen on windows, but never on the linux/bsd systems I use
17:42:09 <crc> I'm still not sure why it happens
17:42:12 <Quartus> It's weird.  Or 'reird' if you prefer.
18:16:46 <Quartus> I'm trying rlwrap under windows, see if the problem is still there.
18:18:53 <Quartus> hee hee hee, I fed it the ANS Forth wordlist and it does command-completion.
18:32:31 <Quartus> I set it up to start up rf with my command-line parameters, send words to a file, and then start it up again with that file as the completion list. :D
18:35:27 --- join: jas2o_ (n=jas2o@wnpp-p-144-134-167-219.prem.tmns.net.au) joined #retro
18:36:04 --- nick: jas2o_ -> jas2o
18:42:25 --- quit: nighty (Read error: 110 (Connection timed out))
18:48:49 <Raystm2> reird bug.
18:48:56 <Quartus> heh
18:49:00 <Raystm2> happend because of comments
18:49:11 <Raystm2> or I should say, stack diagrams.
18:49:15 <Raystm2> as far as I can tell.
18:49:26 <Quartus> Your own personal bug?
18:49:37 <Raystm2> the chess.
18:50:29 <Raystm2> for some still unknown reason, it seems that the stack diagram in one of the mov words was adding values to the stack.
18:50:38 <Raystm2> 18 twice to be more precise.
18:51:03 <Raystm2> I got the but while documenting each word.
18:51:24 <Quartus> ah, the danger of using weird-base numbers.  How did ) slip through? It isn't a defined word, is it?
18:51:39 <Raystm2> hmm.
18:51:52 <Raystm2> don't know that yet. will find out though.
18:52:03 <Quartus> colorforth? retroforth? ans layer?
18:52:12 <Raystm2> retroforth, 
18:52:22 <Raystm2> not ans layer, yet.
18:52:25 <Raystm2> soon.
18:52:38 <Quartus> oh, ok. Words not found sail on by in retroforth without the layer.  Sometimes there's a message about it.
18:52:55 <Raystm2> I see. 
18:52:57 <Quartus> The ANS layer stops: ) <- Exception: undefined word
18:53:29 <Raystm2> good enough reason to start coding to the spec for me.
18:53:51 <Quartus> I would go just a little insane working in a system that didn't stop when I'd misspelled something. :)
18:54:04 <Raystm2> I guess I thought I was working out some mechanics using retro direct. But ANS is where this is going so...
18:54:11 <Quartus> I had trouble enough getting to the point in the ANS layer where it would report errors like that.
18:54:14 --- part: jas2o left #retro
18:54:53 <Raystm2> You did? /me still has much to study. I still havn't read it all.
18:55:09 <Raystm2> I see stuff and I wanna try it and I get distracted. :)
18:55:36 <Quartus> Sure, if I made an error in the file it would generate a whole raft of 'word not found' messages all at once, from wherever the typo was on down, and then crash. Frustrating.
18:55:43 <Raystm2> I try to spend atleast 15 minutes with it every day. Till it is memorized. 
18:56:07 <Quartus> But once I revectored the lookup and installed exception handling, it stopped as it should on my typos. :)
18:56:35 <Raystm2> those are handy indeed.
18:57:12 <Raystm2> I get the impression that crc is working on a new core to make ans completely possible.
18:57:12 <Quartus> I think so, but maybe I'm hopelessly oldschool :)
18:57:16 <Raystm2> :)
18:57:31 <Quartus> He has said he's looking at the remaining bits, which is exciting.
18:57:53 <Raystm2> You don't know how impressed I am with the way you both are working this out. It's exciting to watch.
18:58:07 <Quartus> It's been fun.
18:58:13 <Raystm2> for me as well.
18:58:47 <Raystm2> I think there is a goal out there, and you both see it. aiming for it.
19:01:04 <Quartus> I'm glad to have another command-line accessible standard forth, as I can test my code in multiple compatible environments.  And being faster than gforth-fast doesn't hurt either. :)
19:01:20 <Raystm2> I have to figure out how to get to a working "at-xy" in this windows version.
19:01:32 <Quartus> of retroforth? hmm, not sure you'll have a lot of luck with that.
19:01:35 <Raystm2> Yes. That is cool.
19:01:55 <Quartus> can't you just dump the board as simple ASCII?
19:02:10 <Raystm2> the board is only part of the display. Yes I do that.
19:02:27 <Quartus> crc can confirm, but I don't see any easy way to do at-xy.
19:02:58 <Quartus> Though gforth does it somehow.
19:03:22 <Raystm2> I suppose I will have an array the length of each display possible position and do it that way.
19:03:31 <Raystm2> sorta like the board.
19:03:41 <Quartus> Yes, you could, and repaint the whole screen, but it means having special output words.
19:03:48 <Raystm2> but I can then have lists of last moves and other boards.
19:03:53 <Raystm2> ya;.
19:04:12 <Raystm2> well, I don't see it as much different then the way the board is printing now.
19:04:35 <Raystm2> I just have to treat every line the width of the display with cr's I suppose.
19:04:56 <Raystm2> unless it will wrap, hmmm
19:05:00 <Quartus> rf doesn't have a 'page' command at this point.
19:05:08 <Raystm2> clear.
19:05:21 <Raystm2> or I should say, "clear" is the page command.
19:05:37 * Raystm2 looks up clear.
19:05:56 <Raystm2> it prob'ly calls cls in windows.
19:06:06 <Quartus> It doesn't.  It prints a huge number of blank spaces.
19:06:08 <Raystm2> maybe not, maybe just cr's 25 lines or so.
19:06:13 <Raystm2> ah.
19:06:32 <Raystm2> is that whay page is supposed to do?
19:06:43 <Quartus> which sort-of achieves the same thing, kinda.  Page is supposed to clear the screen and leave you at the top.
19:06:44 <Raystm2> wha/y/t
19:06:51 <Raystm2> ya.
19:07:41 <Raystm2> I have a at-xy that works for crc in *nix's but for me in windows it only works on the same line as the current cursor, won't change lines.
19:09:59 <Quartus> Just looking to see what can be done.
19:10:10 <Raystm2> ooh cool.
19:58:44 --- join: jas2o (n=jas2o@wnpp-p-144-134-167-219.prem.tmns.net.au) joined #retro
20:09:17 --- quit: jas2o ("Chatzilla 0.9.74 [XULRunner 1.8.0.1/2006012608]")
20:19:20 <crc> Raystm2: http://retroforth.net/paste/?id=162
20:19:46 <crc> this was working when my windows installation died
20:27:47 <Quartus> crc, I've been messing with that definition & one I wrote before it, and it occasionlly a) does nothing, b) scrolls to some weird-ass place in the scrollback buffer.
20:28:29 <crc> odd
20:28:31 <Quartus> Does the function-calling thing flush ax to memory before making the call?
20:31:01 <crc> when the call is made, eax contains the address to call; this is replaced by the return value
20:31:32 <Quartus> Ok.
20:31:44 --- part: crc left #retro
20:31:49 --- join: crc (n=crc@pdpc/supporter/active/crc) joined #retro
20:31:49 --- mode: ChanServ set +o crc
20:48:45 <Quartus> Yes, after trying it a few times, it now no longer moves the cursor at all.  Perhaps after a reboot.
20:49:36 <Quartus> maybe it's this rlwrap that's messing with it.
20:50:35 <Quartus> Looks like it!
20:52:21 <Quartus> Ok, this saves a bit of code, as it uses the already-imported words:
20:52:31 <Quartus> : at-xy  16 lshift or kernel32 stdout swap SetConsoleCursorPosition ^ drop ;
20:52:47 <Quartus> That's a Standard Forth name.
22:52:54 <Quartus> A few more Standard words for Windows -- these might be useful to you, Ray: http://quartus.net/retro/facility.fs
22:53:19 <Quartus> I wrote a full-fledged PAGE word, and added get-xy to at-xy.
23:35:14 <Raystm2> Excellent Quartus. I'm sitting here thinking about the ANS re-write.
23:35:50 --- join: Cheery (n=Cheery@a81-197-19-23.elisa-laajakaista.fi) joined #retro
23:35:57 <Quartus> Well, you can clear the screen and set & retrieve the cursor position now.  If you like I can add a word to return size of the currently-visible part of the screen buffer.
23:36:17 <Raystm2> That sounds important. :)
23:36:23 <Raystm2> Please?
23:36:28 <Quartus> Sure, hang on.
23:36:51 <Raystm2> I suppose you base the entire visible display on such a value.
23:45:02 <Quartus> Well, in a console window there's the bit you're looking at, and the bits offscreen.
23:51:21 <Quartus> That should do it.  Reload this and you'll see 'window-size'.  http://quartus.net/retro/facility.fs
23:54:39 <Quartus> Interesting.  If the cursor position is 'down the buffer' such that the scroller is down a bit, window-size reports the number of rows at the cursor.  If you 0 0 at-xy before window-size, it gives the actual size of the window.
23:56:39 <Quartus> Myself, I'd be inclined to code to the visible area, to a minimum of 80x25.
23:57:27 <Quartus> You can get a console window with 40 chars across, but I don't know why anyone would want one. :)
23:59:58 <Quartus> If you have a hard dependency on a non-scrolling console of a certain size, you can request one ahead of time with a mode con: command, or I'm sure there's an api call to make it happen.
23:59:59 --- log: ended retro/06.08.26