00:00:00 --- log: started forth/04.08.10
00:01:03 <arke> oberon looks like a more modern BASIC
00:01:05 * arke runs
00:01:07 <arke> ^_^
00:02:26 <kc5tja> Oberon and Python are both modeled after Modula-2.
00:09:22 --- quit: mur_ (Read error: 110 (Connection timed out))
00:50:46 <arke> wow.
00:50:58 <arke> steve jobs was a real asshole to jef raskin :/
00:54:04 <kc5tja> Yep.
00:54:42 <arke> hmm
00:55:02 <arke> steve jobs is one of those super-geniuses that know it :/
00:56:51 <arke> woah
00:56:52 <arke> :
00:57:43 <kc5tja> No, he's just one of those super-eccentric CEOs who's very successful at making seemingly impossible ventures work.  And he knows it.  :)
00:58:05 <arke> http://folklore.org/StoryView.py?project=Macintosh&story=Tell_Adam_Hes_An_Asshole.txt&sortOrder=Sort%20by%20Date
00:58:17 <arke> heh :)
00:58:36 <arke> well, from the stuff ive read from this site, i'd say both definitions are equally true
00:58:48 --- join: lyca (alycat@rasterburn.org) joined #forth
01:33:27 <kc5tja> OK, bedtime for me.
01:33:36 --- quit: kc5tja ("THX QSO ES 73 DE KC5TJA/6 CL ES QRT AR SK")
02:09:56 --- join: Topaz (jonny@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
03:55:11 --- join: sallust_ (~reynaert@205.111-136-217.adsl.skynet.be) joined #forth
04:00:59 --- quit: sallust (Read error: 110 (Connection timed out))
04:08:01 --- join: crc (crc@0-1pool52-21.nas39.philadelphia1.pa.us.da.qwest.net) joined #forth
04:17:08 --- join: mur_ (~mur@kyberias.uiah.fi) joined #forth
04:28:42 --- quit: mur (Read error: 110 (Connection timed out))
04:32:26 --- quit: crc ("Time for bed... Goodnight all!")
05:20:47 --- join: mur (~mur@smtp.uiah.fi) joined #forth
05:23:43 --- quit: mur_ (Read error: 60 (Operation timed out))
05:41:59 --- quit: tgunr (Read error: 104 (Connection reset by peer))
05:42:41 --- join: tgunr (~davec@148.63.4.107) joined #forth
06:14:23 --- quit: tgunr (Read error: 104 (Connection reset by peer))
06:15:50 --- join: tgunr (~davec@vsat-148-63-4-107.c001.g4.mrt.starband.net) joined #forth
06:37:01 --- join: mur_ (~mur@uiah.fi) joined #forth
06:38:10 --- quit: mur (Remote closed the connection)
07:20:13 --- join: mur (~mur@kyberias.uiah.fi) joined #forth
07:22:32 --- quit: mur_ (Read error: 60 (Operation timed out))
07:55:54 --- join: warpzero (~warpzero@mi209.dn189.umontana.edu) joined #forth
08:10:55 --- join: I440r_ (~mark4@216-110-82-203.gen.twtelecom.net) joined #forth
08:15:15 <warpzero> needs more I440rs
08:15:21 <warpzero> 3 is just not enough'
08:23:49 --- quit: Topaz ("Leaving")
08:27:33 --- quit: I4404__ (Read error: 110 (Connection timed out))
08:30:39 --- quit: tgunr (Read error: 104 (Connection reset by peer))
08:32:20 --- join: tgunr (~davec@vsat-148-63-4-107.c001.g4.mrt.starband.net) joined #forth
08:46:53 --- quit: tgunr (Read error: 104 (Connection reset by peer))
08:48:43 --- join: tgunr (~davec@vsat-148-63-4-107.c001.g4.mrt.starband.net) joined #forth
09:05:03 --- join: mur_ (~mur@kyberias.uiah.fi) joined #forth
09:06:56 --- quit: mur (Read error: 60 (Operation timed out))
09:26:01 --- join: mur (~mur@uiah.fi) joined #forth
09:37:27 --- quit: mur_ (Read error: 110 (Connection timed out))
10:11:22 --- join: kc5tja (~kc5tja@66-74-218-202.san.rr.com) joined #forth
10:11:31 --- mode: ChanServ set +o kc5tja
10:11:46 <kc5tja> WELL NOW....looks like I will *NOT* be using FPGAs in my system design.
10:12:24 <kc5tja> At $60 to $104 for a *small* chip in single quantities, that'd put the cost of a Kestrel squarely at $400 or so ($200 for raw manufacture, plus 100% profit).
10:12:54 <Robert> Shit, are they that expensive?
10:13:02 <Robert> I thought they were a lot cheaper.
10:13:19 <kc5tja> In *single* quantities.
10:13:26 <kc5tja> They are cheaper when you get into volume production.
10:14:06 <kc5tja> I've sent an e-mail to an FAE (Field Application Engineer) about what I'm looking to do, and he's expected to get back to me on more concrete pricing, but it is not looking too good initially.
10:14:54 <kc5tja> Therefore, (a) I'm thinking of going back to a pre-existing CPU architecture, and (b) I'm going back to having the CPU feed the video directly instead of a video chip.
10:15:35 <kc5tja> There may be an opportunity that a CPLD might work for my needs, but doggone it, they are damn limited compared to an FPGA.
10:16:44 <Robert> :/
10:18:17 <kc5tja> The FPGA solution is still doable if I can find a processor that runs at 100MHz or more and comes with a kick-butt operating system for it (something that is at least AmigaOS-class).
10:18:45 <kc5tja> The OS I can handle.  It's the CPU that I won't be able to.  :)  And the SDRAM.
10:20:25 --- quit: tgunr (Excess Flood)
10:20:50 <kc5tja> I guess this brings me right back to using the 65816 microprocessor.
10:21:23 <Robert> Lots of going back and forth... But I guess you can't be blaimed,
10:21:39 <kc5tja> Well, it's all because these technologies are unnecessarily exotic.
10:21:52 <kc5tja> I mean, there really is no reason why an FPGA should cost that much.
10:22:11 --- join: tgunr (~davec@vsat-148-63-4-107.c001.g4.mrt.starband.net) joined #forth
10:22:23 <kc5tja> The XCV100 has been available now for, what, 5 years?  It has sold like *hotcakes*.  Yet it's still priced as if it were a brand-new product.
10:22:33 <kc5tja> At the very least, they should have priced it at $25.
10:22:38 <kc5tja> I can live with $25.
10:22:45 <kc5tja> $105 is clean out of the question though.
10:23:05 <kc5tja> Oh, and the smaller unit, the XCV100 144-pin variety, costs even more: $167!
10:23:15 <kc5tja> $105 for 256 pins, $167 for 144.  Explain THAT one to me.
10:23:18 * kc5tja sighs
10:24:19 <kc5tja> Even if I could pack the entire Kestrel onto one chip, that still means the Kestrel would cost a minimum of $200 to $300 for you folks.
10:24:32 <kc5tja> It's rediculous.
10:24:51 <kc5tja> I really have to wonder how these other independent computer companies are able to sell their products for as little as $250.  I really do.
10:27:39 <Robert> Bribes, probably.
10:28:23 <kc5tja> Well, needless to say, I am very, very disappointed.
10:28:45 <kc5tja> If it's not one thing, it's another, and I'm 75% to the point where I'm just about to give up on the whole fucking project.
10:29:33 <kc5tja> I mean, here was a NICE opportunity to have some fun, and earn a little side income too.
10:29:36 <kc5tja> But no, I can't do it.
10:29:59 <kc5tja> The price point of the Kestrel was already increased to $130 once, and people still hemmed and hawed about that.
10:30:06 <kc5tja> Imagine if I said, "Oh, btw, now it's $450."
10:30:42 <kc5tja> Nobody in their right mind would spend $500 on a computer that ran, tops, at 25MHz, and sported only 640x480, 256-color graphics, with only two-voice stereo sound.
10:30:46 <kc5tja> Absolutely nobody.
10:32:43 <Robert> Maybe you could find someone who has been in a coma since 1990... but I agree, it's not very realistic.
10:37:40 <kc5tja> Dude, a Spartan IIe FPGA costs about $500 in its biggest form.   >:(  That sucks ASS.
10:38:00 <kc5tja> $125 in its smallest form.
10:38:13 <kc5tja> Definitely cheaper than the Virtex, but still... 
10:38:55 <Robert> Heh.
10:39:06 <Robert> Are they expensive to produce, or just terribly over-priced?
10:39:15 <kc5tja> They're fucking way over-priced.
10:39:52 <kc5tja> There are only two companies making FPGAs today: Xilinx and Altera.
10:39:55 <Robert> But are they any hard at all to produce?
10:40:13 <kc5tja> No.  They are about as hard to produce as giant RAM chips.
10:40:24 <kc5tja> Because that's more or less all they are.
10:40:32 <kc5tja> RAM with a bunch of spare logic on the board.
10:40:40 <kc5tja> s/board/chip/
10:42:37 <kc5tja> Anyway, because there are only two companies producing FPGAs, they can command a premium price, even for chips with as little as 100K gates on them.
10:43:29 <Robert> Not very nice of them.
10:44:02 <Robert> But I guess, that's what happens when they realize nobody (except the majority - customers) would win on lowering prices.
10:45:01 <kc5tja> Well, I either need to find a way to work around this, or just terminate the project on the grounds that it is just plain too expensive.
10:45:36 <Robert> Weren't you able to design a pretty cheap solution with a normal microprocessor?
10:46:06 <kc5tja> The CPU did *everything*, including refreshing the display.
10:46:22 <Robert> Hehe
10:46:42 <kc5tja> But the problem with that system was that I could purchase RAM chips fast enough to keep up with the CPU.
10:46:54 <kc5tja> s/could/could not/
10:48:58 <Robert> Oh. :/
10:53:57 --- join: mur_ (~mur@mgw2.uiah.fi) joined #forth
10:54:04 <kc5tja> I mean, the final decision on this will come when I hear back from the FAE.
10:54:39 <kc5tja> AH HA!
10:54:47 <Robert> Hm?
10:55:21 <kc5tja> I think I have found something more usable: An XCS20 Spartan-I FPGA, 208 pins, for $40.40.  Not a whole lot of room on the chip to fit a big-name CPU, so I might still have to go with a commodity CPU like MIPS or ARM or something.
10:55:45 <kc5tja> But it should be big enough to fit the I/O processor and peripherals on (e.g., video, maybe some audio, UART, etc.)
10:57:02 <Robert> Cool
10:58:46 <kc5tja> But then there are the costs of RAM too.
10:59:35 <kc5tja> With a system bus running a 6.3MHz, I can use asynchronous SRAMs for system memory, but SRAMs are pretty expensive.  I suspect about $15 to $25 a pop.
10:59:42 <kc5tja> I'd have to do the research on that though.
11:01:32 <kc5tja> I don't know.
11:01:34 <kc5tja> I have to think about it.
11:04:52 <kc5tja> Maybe if I up'ed the CPU speed to somewhere in the 100MHz to 150MHz region, the $350 wholesale price would be worth it to some people.
11:05:16 --- quit: mur (Read error: 110 (Connection timed out))
11:05:45 <kc5tja> But for that, I had better provide a real OS for it, and not just Forth In ROM.
11:06:18 <kc5tja> And that means open source OS development, and that means Dolphin.
11:06:24 <kc5tja> Which I might as well develop for the x86 anyway.
11:06:44 <kc5tja> (sorry, I'm just thinking "out loud" to myself.  If I get annoying, just tell me to shut up.)
11:06:47 <kc5tja> :)
11:10:03 <Robert> I'm doing various other things, so I can't reply instantly. I think a 'real OS' could be nice, but a good-enough Forth with some programs would also be fine.
11:10:26 <kc5tja> I'm talking about the salability of the product.
11:10:41 <kc5tja> Sure, a Forth-only system would be OK for you, since you are a Forth programmer and entheusiast.
11:11:13 <kc5tja> But to someone looking on my website, just browsing, they'd be shocked to find no conventional OS was available for it.  That would basically mean the hardware is over-priced for what they are getting in return.
11:11:42 <Robert> I'm sure everyone could learn the basic Forth words to load programs.
11:11:56 <Robert> I mean, lots of C64 users didn't know BASIC...
11:12:31 <kc5tja> True.
11:18:06 --- quit: tgunr (Read error: 104 (Connection reset by peer))
11:19:50 --- join: tgunr (~davec@vsat-148-63-4-107.c001.g4.mrt.starband.net) joined #forth
11:43:28 --- quit: tgunr (Read error: 104 (Connection reset by peer))
11:45:12 --- join: tgunr (~davec@vsat-148-63-4-107.c001.g4.mrt.starband.net) joined #forth
11:49:46 <madwork> load" myprogram.f"
11:49:46 <madwork> run
11:53:58 <kc5tja> guys, I just don't think this will happen.
11:53:59 <kc5tja> Sorry.
11:54:18 <kc5tja> It's looking to just be getting more and more expensive with each design iteration, and I KNOW that it's not getting any more complex.
11:54:48 <kc5tja> I was just looking at a PowerPC 405CR embedded processor.  Has lots of stuff on-board, but it's a 316-pin BGA.  Yuck.
11:54:57 <kc5tja> I don't even have the facilities to solder that thing onto the board!
11:55:21 <Robert> Heh.
12:01:34 <I440r_> bga ?
12:01:37 <I440r_> ball ?
12:05:36 --- join: mur (~mur@mgw2.uiah.fi) joined #forth
12:07:59 <kc5tja> Ball Grid Array, yes.
12:09:43 --- join: Topaz (jonny@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
12:14:13 <kc5tja> Maybe what I should do *IS* implement that virtual machine concept, and just sell the simple system with a really low-end processor card
12:14:39 <kc5tja> Then, if people want a higher-end unit, all they need to do is purchase a higher-end (or build their own) processor card.
12:15:51 <madwork> Sounds better than giving up completely.
12:15:57 <arke> hi1
12:16:00 * arke woke up
12:16:26 <kc5tja> madwork: Yeah, but it just wreaks of TI-99/4A.  :/
12:16:47 <madwork> It also reeks of "finished product."
12:17:01 <I440r_> try soldering pins to the bga hehehe
12:17:34 <kc5tja> I440r_: You'll damage the chip.  The solder balls exist right on the silicon wafer itself -- so the whole wafer has to be heated uniformly.
12:17:58 --- quit: mur_ (Read error: 110 (Connection timed out))
12:19:14 <kc5tja> madwork: The problem is, the 65816 isn't going to be terribly all that great at emulating another instruction set.
12:19:28 <I440r_> ouch
12:19:45 <kc5tja> madwork: Well, let me rephrase.  It won't interpret the virtual instruction set.  It will on-the-fly translate the code on an as-needed basis.
12:20:12 <kc5tja> madwork: But with the CPU being speed limited (say, 8MHz), it's not going to run all that great.
12:20:21 <Topaz> considered an ARM or something?
12:20:27 <Topaz> (if you're going to use a drop-in proc)
12:20:33 <kc5tja> ARM CPUs are cheap.  The *SUPPORT* for those isn't.
12:20:38 <arke> folklore.org is a great site
12:20:38 <madwork> So then, code directly for the 65816.
12:20:54 <Topaz> support as in C compiler?
12:21:07 <kc5tja> Topaz: Support as in bus controller, memory interfaces, etc.
12:21:12 <Topaz> ah, ok
12:21:49 <madwork> Why not dumpster dive for old keyboards, crack them open and take the *free* processors out of them. Code for those. Maybe run them in parallel for fun.
12:22:10 <kc5tja> madwork: umm....you ARE aware that the original intent was to sell this as a kit right?
12:22:23 <kc5tja> I am not going to dumpster dive for a living.
12:23:34 <imaginator> Dumpster diving doesn't sound bad to me.  I wonder what good places are to go diving.
12:23:34 <madwork> You'd not have to once you got the project going.
12:23:59 <Topaz> hmm, i didn't realise that peripheral chips were a significant part of the cost
12:24:01 <madwork> There's a dumpster behind work here that's often got hardware thrown into it... PCs, monitors, keyboards, mice, etc.
12:24:41 --- join: slava (~slava@CPE00096ba44261-CM000e5cdfda14.cpe.net.cable.rogers.com) joined #forth
12:24:44 <kc5tja> madwork: I invite you to check out what they have in them.  All systems on chips, which in and of themselves are OK for what they're intended for.
12:25:08 <kc5tja> Clock for clock, an 8051 (well, 8048) chip or its clones have worse performance than even a 6502.
12:25:28 <madwork> 8048, 8042...
12:25:59 <madwork> And a Kestrel with no processor at all has even worse performance. :P I just want to see a prototype at this point! I don't even care what it is anymore.
12:26:42 <madwork> BTW... what's a good alternative Forth name for "type" as in datatype... or should I use "print" instead of "type" and use "type" for getting a value's type?
12:27:04 <madwork> Or maybe >type
12:27:28 <kc5tja> madwork: The issue isn't the prototype.  The issue is the finished production cost.  I've been told repeatedly that if the cost of the unit exceeds $100 to $130, they wouldn't want one.  Ergo, if I'm going to spend more than that on producing a kit, I may as well just invest in the FPGAs and build for myself and that's it.
12:28:11 <arke> madwork: typeof maybe
12:28:33 <kc5tja> Why not just "datatype"?
12:29:40 --- join: crc (crc@0-1pool148-93.nas63.philadelphia1.pa.us.da.qwest.net) joined #forth
12:30:07 <madwork> or perhaps "valuetype". I don't really refer to "data" anywhere in the API.
12:30:07 <slava> madwork, i have a word type-of
12:30:25 <madwork> I think I'll just go with >type and $>type
12:30:25 <kc5tja> How about "kind"?
12:30:28 <slava> and one word for each type like string? vector? number? etc.
12:30:30 <madwork> Hmm.
12:30:34 <slava> or class
12:31:34 <Topaz> there's always those hyped-up dallas 8051s that can manage 33 MIPS
12:31:39 <Topaz> (and they give out samples)
12:31:47 * kc5tja vomits
12:31:51 <Topaz> ;)
12:31:55 <kc5tja> I won't touch an 8051 with a 10 foot pole.
12:32:05 <kc5tja> I'll use a Z-80 before I use an 8051.
12:32:06 <madwork> Not even to smash it?
12:32:43 <arke> my calculator has a z80.
12:32:47 <arke> hmm....
12:33:00 <arke> yep 8051 is teh suck.
12:33:03 <kc5tja> Most calculators today have Z-80s in them.
12:33:12 <arke> z80 shouldnt be _that_ bad though ... :/
12:33:39 <Topaz> if only they made AVR processors ;)
12:35:56 <Topaz> would a microcontroller with enough storage work? i guess it's a kludge, but some do have huge onboard memories
12:37:36 <kc5tja> Wow.
12:37:56 <kc5tja> MIPS RM5231A (64-bit CPU, 32-bit external bus, 32K caches for both instruction and data, superscalar): $22.50
12:37:59 <kc5tja> For *1*.
12:38:02 <Topaz> oooooooh
12:38:11 <Topaz> mewants
12:38:18 <Topaz> how fast?
12:38:28 <kc5tja> 300MHz.
12:38:31 <Topaz> !
12:38:33 <kc5tja> (as far as I can see)
12:38:33 <Topaz> (does it have an MMU, incidently?)
12:38:36 <kc5tja> Yes.
12:38:42 <kc5tja> Software managed TLBs for memory management.
12:38:44 <Topaz> hah, you could chuck linux at it
12:38:47 <Topaz> cracking
12:38:52 <kc5tja> Linux has been ported to it.
12:39:09 <kc5tja> But dude, *superscalar*.
12:39:10 <arke> hmmmmm
12:39:12 <Topaz> i guess those things power every wireless access point and mp3 player available
12:39:12 <kc5tja> 64-bit.
12:39:19 <crc> What's the power draw?
12:39:20 <arke> kc5tja: so you could slap one of those MIPSes in there?
12:39:27 <kc5tja> crc: According to the datasheet, 1W.
12:39:38 <arke> wait --- superscalar! :)
12:39:42 <arke> kc5tja: so you could slap one of those MIPSes in there?
12:40:00 <Topaz> heh, was about to say that you could get 66mhz atmel AT91 ARM Thumb microcontrollers with an external bus interface for $22, but i think you beat me ;)
12:40:03 * slava is facing a dilemma
12:40:19 <kc5tja> arke: I have to do more research.  This could be a case where the CPU is cheap but support logic isn't type thing.  But hell, even if I have to hack something up in discrete component logic, that's still **DAMN** cheap for that kind of CPU power!
12:40:23 <slava> should i use guard pages to check for stack overflow/underflow, and cut off cfactor from mmu-less targets?
12:40:34 <Topaz> you could make a supercomputer out of those things, mmmm
12:40:40 <arke> kc5tja: that would be AWESOME! :)
12:40:40 <slava> or should i stick with the current 'check after each primitive' which has a 2x performance hit?
12:40:41 <kc5tja> Topaz: Yes, you could.
12:40:47 <Topaz> imagine, 2000 of them consuming less than a kettle
12:41:01 <arke> kc5tja: i would pay a bit more for a kludge to make that work if its 300mhz superscalar
12:41:11 <kc5tja> slava: You could always make it a compile-time option.
12:41:16 <arke> kc5tja: thats like a 700mhz intel processor :D
12:41:31 <kc5tja> arke: Remember that Intel's are also superscalar.
12:41:59 <kc5tja> This thing would have the power of a Pentium II or so.
12:42:09 <Topaz> has a 64-bit core though, by the look of it
12:42:10 <kc5tja> And plus it's 64-bit as well.
12:42:12 <arke> massively awesome.
12:42:14 <arke> massively awesome.
12:42:14 <arke> massively awesome.
12:42:20 <slava> kc5tja, true
12:43:06 <Topaz> where you getting such chips from then?
12:43:16 <Topaz> (i want one ;)
12:43:24 <kc5tja> I would have to get them from http://www.insight-electronics.com .
12:43:40 <kc5tja> HEY...WAIT A MINUTE!!!
12:43:50 <kc5tja> The 5231A is the chip I used to program back when I was working for Hifn!
12:43:56 * kc5tja thought that chip looked awfully familiar!!
12:43:58 <Topaz> haha
12:44:57 <arke> :)
12:45:11 <arke> so you already know it pretty well?
12:45:31 <kc5tja> Hmm....and MIPS instruction set is relatively easy to emulate in software too.  I could probably hack on SPIM (the emulator for Linux) to make it work for the Kestrel if that were the case.
12:45:39 <kc5tja> The only issue is that 125MHz CPU bus.  :D
12:46:12 <Topaz> is the core static? (will it copy with being heineously underclocked? ;)
12:46:38 <kc5tja> I don't know.
12:46:46 <kc5tja> It's likely that it will be static.
12:46:51 <kc5tja> But I can't guarantee it
12:47:08 <arke> kc5tja: :)
12:47:16 <arke> kc5tja: why is the bus an issue?
12:47:48 <kc5tja> Because at those frequencies, even the shortest wire is an antenna.
12:47:54 <kc5tja> And you have 32 to 64 antennas on the board.
12:48:01 <kc5tja> If one transmits, the others receive.
12:48:12 <kc5tja> Therefore, you get a LOT in cross-talk and interference.
12:48:30 <kc5tja> Plus, it's just plain damn hard to find logic that will run that fast.
12:48:49 <kc5tja> It'll even be pushing an FPGA to its limits.
12:48:51 <Topaz> have you considered CPLDs and stuff for glue-logic? they're cheaper than FPGAs
12:49:00 <kc5tja> Yes.
12:49:04 <Topaz> (though i'm not entirely sure what i'm talking about, i've never used one :)
12:49:10 <kc5tja> BUT...I'm not sure I can get an SDRAM controller into a CPLD.
12:49:46 <Topaz> ah, hmm
12:49:50 <Topaz> i guess you have to use 'real ram' ;)
12:50:11 * Topaz pats his AT90S1200 microcontrollers with no ram whatsoever
12:50:11 <kc5tja> Topaz: Can you find me "real RAM" that will function at 100MHZ?  50MHz?  Heck, even 25MHz?
12:50:14 <kc5tja> Good luck.
12:51:06 <Topaz> heh, as in RAM with real-world interfaces instead of fantasy-land A/D/R/W/EN ;)
12:51:27 <kc5tja> Every asynchronous RAM chip I'm aware of only has those signals.
12:51:33 <kc5tja> So that's about as real-world as it gets.
12:52:01 <Topaz> but yeah, you've unfortunately got to associate with SDRAM
12:52:10 <Topaz> (about which i know nothing, admittedly)
12:52:36 <kc5tja> Well, for some, it can take up to eight clock cycles to write one word to memory.
12:52:57 <Topaz> but they're designed for bursts, right?
12:53:16 <kc5tja> yes
12:53:44 <kc5tja> I wonder what would happen if I drove the CPU at 300MHz internally, but slapped it on a 6MHz bus.
12:53:53 <Topaz> oh, so you've got to worry about cache controllers and stuff
12:53:58 <Topaz> hmm, that's quite a multiplier :)
12:54:04 <kc5tja> CPU has internal 32K instruction and data caches.
12:54:09 <kc5tja> I don't need to worry about those.
12:54:19 <Topaz> i guess a major upside of FORTH is its ability to fit in small caches
12:54:29 <kc5tja> I drive the CPU at 100MHz to 125MHz, but the external bus it attaches to would be driven at 6MHz.
12:55:05 <Topaz> so filling the cache would take ages, but it'd scream at repetitive-data-accesses?
12:55:19 <kc5tja> ONLY if those accesses were in the cache, yes.
12:55:32 <kc5tja> (AND assuming the caches were configured for write-back and not write-through)
12:56:14 <warpzero> hey kc5
12:56:16 <warpzero> howsit
12:57:29 <kc5tja> warpzero: Read the backlog to read my latest dilemma.
12:59:57 <warpzero> oh huh
13:00:06 <warpzero> ive used SPIM
13:01:10 --- join: mur_ (~mur@kyberias.uiah.fi) joined #forth
13:01:57 <kc5tja> Topaz: Where can I find more information about the AT91s you were talking about?
13:02:01 * kc5tja just wants to compare.
13:02:30 --- quit: mur (Read error: 60 (Operation timed out))
13:02:42 <warpzero> [13:39:10] <kc5tja> But dude, *superscalar*.
13:02:50 <warpzero> thats going in the quotebook
13:03:00 <Topaz> sec
13:03:12 <Topaz> http://www.atmel.com/dyn/products/param_table.asp?family_id=605&OrderBy=part_no&Direction=ASC
13:03:50 <Topaz> digikey has the AT91R40008 for $22, only one i checked
13:04:10 <Topaz> it has an outrageous number of pins, if that's an advantage ;)
13:04:13 <Topaz> (think of all the LEDs!)
13:04:54 <warpzero> i want
13:05:09 <warpzero> lots of pins make a warpzero happy
13:05:11 <Topaz> interesting, has 256kb of SRAM as the only internal storage, and loads everything off external memory
13:05:24 <Topaz> rather uP-like for a uC
13:06:23 <Topaz> ooh, you can clock it at 0hz!
13:06:24 <Topaz> fun fun
13:06:48 <kc5tja> Topaz: It's a BGA.  No, definitely not an advantag.e
13:06:59 <arke> 0hz? lol
13:07:01 <Topaz> hmm, they did TQFP ones, i'm sure
13:07:13 <arke> frozen in time :D
13:07:19 <Topaz> this one is TQFP100, but i didn't check on digikey, hmm
13:07:29 <Topaz> http://www.digikey.com/scripts/DkSearch/dksus.dll?Criteria?Ref=153977&Site=US&Cat=31523696
13:07:32 <Topaz> TQFP-100
13:09:10 <Topaz> ooh, here's a rather over-capable chip: http://www.digikey.com/scripts/DkSearch/dksus.dll?Detail?Ref=155886&Row=206143&Site=US
13:09:14 <Topaz> $27 dollars though
13:09:17 <Topaz> PFQP package
13:09:30 <Topaz> 147 IO lines, you know you want ;)
13:09:45 <kc5tja> This chip is doable, if you don't mind CPU-fed video.
13:09:49 <Topaz> wait, 196 IO lines! :D
13:09:59 <kc5tja> Since the 256K of RAM apparently has no support for external bus-master access.
13:10:05 <Topaz> onboard USB host and USB device, weird
13:10:06 <Topaz> hmm
13:10:21 <kc5tja> On-board USB *host*?
13:10:24 <Topaz> yeah
13:10:29 <kc5tja> THAT may make it compelling.
13:10:35 <kc5tja> What is the part number?
13:10:38 <kc5tja> Atmel part number?
13:10:50 <Topaz> AT91RM9200
13:11:00 <Topaz> ooh, 200 MIPS
13:11:19 <Topaz> this has some really silly features
13:11:22 <Topaz> onboard ethernet MAC!
13:11:28 <Topaz> 100BaseT!
13:11:39 <kc5tja> Yeah, that is silly.
13:11:41 <kc5tja> Who would want that?
13:11:51 <Topaz> :D
13:12:08 <kc5tja> How much was that chip?
13:12:17 * kc5tja can't even see the chip yet in Atmel's webpage
13:12:20 <arke> i could plug it into my eth bridge and IRC from it! :)
13:12:20 <Topaz> 27 dollars
13:12:29 <Topaz> has onboard SDRAM controller, apparently
13:12:43 <Topaz> (but still the issue of being CPU-internal)
13:12:54 <Topaz> actually, this thing has ROM
13:13:08 <Topaz> atmel page: http://www.atmel.com/dyn/products/product_card.asp?part_id=2983
13:13:20 <Topaz> heh, i feel like buying one of these to play with
13:14:44 <Topaz> "Glueless connection to CompactFlash"
13:14:45 <Topaz> that's kinda fun
13:15:17 <Topaz> hmm, this chip does everything
13:15:23 --- join: skittle (XINU@12-222-128-22.client.insightBB.com) joined #forth
13:16:00 <kc5tja> Yeah.
13:16:06 * kc5tja might consider using this as the core.
13:16:18 <arke> :)
13:16:24 <arke> ethernet makes me happy
13:16:27 <Topaz> '? Four 32-bit PIO Controllers with Up to 122 Programmable I/O Lines, Input Change
13:16:36 <Topaz> Interrupt and Open-drain Capability on Each Line
13:16:43 <Topaz> (sounds like fun)
13:16:45 <kc5tja> The ROM -- is that EEPROM inside the chip?
13:16:48 <Topaz> yeah
13:16:52 <Topaz> JTAG-programmable i think
13:17:07 <Topaz> at least, atmel's AVRs have onboard flash, and are outrageously easy to program
13:17:22 <Topaz> actually, no, it says 'ROM'
13:17:23 <Topaz> :/
13:17:28 <Topaz> no mention of any flash
13:18:13 <Topaz> that might be an issue, but it mentions something about a default boot program
13:19:35 <kc5tja> Well, it can boot from external memory, I hope?
13:20:46 <Topaz> yeah, the development board appears to
13:20:57 <Topaz> although the internal bootrom plays a part in it apparently, appears to bootstrap it
13:21:04 <Topaz> (might come pre-programmed i think)
13:21:30 <Topaz> oh, wait, no, ignore that
13:21:45 <Topaz> you can choose whether to boot from internal ROM or external flash with a jumper
13:21:53 <Topaz> (BMS pin)
13:22:59 <kc5tja> OK, the problem is exploiting Ethernet -- that interface is 100% foreign to me.
13:23:05 <kc5tja> OK, I have a MAC -- now what?
13:23:24 <warpzero> now we sing
13:24:56 <Topaz> atmel documentation is usually rather good, i'm looking now
13:26:11 <warpzero> data entry on windows sucks
13:26:21 <warpzero> no middle-click copy
13:27:30 <Topaz> you have to use DMA it appears
13:28:35 <Topaz> you simply DMA from memory into a FIFO in the MAC, and the frame is encapsulated and sent automatically, it appears
13:32:02 <Topaz> (you set your MAC in a seperate register, which determines whether received frames are kept, if so they end up in a 1.5k buffer in RAM (which you set up beforehand)
13:32:53 <Topaz> i'd say it looks easy, although there's a number of things to be initialised first
13:33:04 <Topaz> like the 'AIC' (advanced interrupt controller)
13:37:07 * Topaz demonstrates his stunning datasheet-reading ability ;)
13:37:25 <Topaz> i love the number of acronyms in this
13:37:57 <kc5tja> Topaz: What I meant was the MAC interface still needs to talk to a PHY interface.
13:38:13 <Topaz> hmm, oh
13:39:14 <Topaz> (heh, draws 31.7mA at 1.8/3.3V it appears, talk about low-power)
13:39:27 --- quit: skittle (Remote closed the connection)
13:39:38 <Topaz> amusingly the mac uses the most current
13:40:17 <Topaz> i see, yeah, it doesn't really elaborate on the PHY itself :/
13:44:23 --- join: mur (~mur@mgw2.uiah.fi) joined #forth
13:45:15 <Topaz> http://www.asystelectronic.si/isystem/OnChip/ARM/ITAT91RM9200_V11.pdf
13:45:27 <Topaz> has the schematic of a dev board at the end, ethernet PHY included
13:46:08 <Topaz> (look at all the decoupling! ;)
13:46:27 <Topaz> (is that 50 100n capacitors i count?)
13:46:36 <Robert> Heh.
13:47:40 <Topaz> hmm, laying out that PCB must be a bit of a task
13:54:29 <Topaz> atmel have the entire schematic of their dev board around somewhere too, apparently
13:54:54 <Topaz> heh, on the product page, *slaps self*
13:55:52 <Topaz> ooh, and some uni project, even with PCB layouts!
13:55:54 * Topaz wants to make
13:56:26 <kc5tja> I see; the boot ROM contains a bootloader which I can use to make the Forth system in, and use the internal 16K RAM as the OS bootstrap memory.
13:56:38 <kc5tja> Then external RAM can be configured via the memory controller and EBI interfaces.
13:57:45 <kc5tja> Although slower than the MIPS, I think if I were to market a computer kit, this would probably be the one to start off with.
13:59:51 <Topaz> there's always the novelty of running at 1hz
14:00:02 <Topaz> (and the huge simplification of PCB design ;)
14:00:14 <arke> :
14:00:18 <arke> :)
14:00:47 <Topaz> well, linux runs on this chip, apparently
14:00:58 <Topaz> (if you want overkill :D)
14:01:49 <arke> :)
14:04:24 --- quit: mur_ (Read error: 110 (Connection timed out))
14:06:41 <kc5tja> Man, I have to think about it.
14:07:11 <Topaz> i don't think i'll go near any heavy-duty processors in the near future, heh...
14:07:13 <Topaz> too many pins!
14:07:23 * Topaz pats his 8-pin AVR
14:07:32 <Robert> Hehe :)
14:07:49 <Robert> That's the spirit, 8 pins, 8 bits, 8 stack items.
14:07:53 <Robert> The perfect Forther.
14:07:55 <Topaz> sure it's only 8 MIPS, and only has 6 usable IO lines ;)
14:08:28 <Topaz> though 8 MIPS for a barely-visible black square thing is fairly stunning
14:08:46 <Topaz> (it has a 6-channel 10-bit ADC onboard, the ridiculousness)
14:09:14 --- quit: crc (Client Quit)
14:09:40 <Robert> Hehe, sounds like this PIC12F675s I got.
14:10:21 * Topaz hunts through his electronics box
14:10:25 <Topaz> hrm, it only has 7 pins now :/
14:10:32 <Robert> Heh.
14:10:39 <Topaz> and, of course, the negative power pin fell off
14:10:54 <kc5tja> I need some reasonably heavy-duty processors (note I said "reasonably") because people want a machine they can run Forth on, and because PCs today are just plain too complex to grok in their entirety.
14:10:58 <Robert> Hm, "negative power". :)
14:11:10 <Topaz> ok, 'ground' :P
14:11:25 <Topaz> kc5tja: yeah, would be the way to go i guess
14:12:23 <Robert> If I didn't used sound and video so much, I could do with pretty much any computer.
14:12:50 --- join: crc (crc@0-1pool32-109.nas34.philadelphia1.pa.us.da.qwest.net) joined #forth
14:12:58 <Robert> IRC is the #1 priority, web browsing #2.
14:13:11 <Robert> Oh, and programming tools.
14:13:31 <kc5tja> Robert: Interestingly enough, I have similar requirements.  :D  Although e-mail is #1 for me, browsing #2, and IRC is #3.
14:13:40 <Robert> Hehe
14:13:58 <arke> :)
14:14:15 <Robert> Depends on the mood. If I'm this bored, IRC and web browsing are up there. Sometimes coding is everything, sometimes I use the mail account a lot.
14:14:34 <Robert> Oh, and some days I even use games!
14:14:36 --- part: rpc left #forth
14:14:37 <kc5tja> Well, I'm going to take a break from the Kestrel for a bit.
14:14:43 <kc5tja> Maybe won't get back to it for another week or so.
14:14:46 <Robert> Torturing my brain with.
14:14:49 * kc5tja is going to let all this stuff simmer for a while.
14:15:27 * crc is considering ways to implement 64bit integers in RetroForth
14:15:44 <Robert> As the default size?
14:16:43 <crc> No
14:16:50 <crc> As an optional add-on :-)
14:17:02 <I440r_> is 32 bit your default size ?
14:17:08 <crc> yes
14:26:15 * crc is also planning to do an ARM port later this year
14:27:36 <I440r_> you can look at isforths double number stuff, that might help you
14:27:59 <I440r_> thats all 64 bit stuff - no big deal to implement
14:29:50 <Robert> Oooh, more code to steal. Thanks for pointing that out, I440r_ 
14:30:40 <arke> Robert: for F2 or something else?
14:31:00 <Robert> I don't know.
14:31:05 <Robert> Could be useful
14:31:13 <Robert> Just like his DNS code...oh, wait...
14:31:38 <arke> hahahaha
14:31:51 <arke> :)
14:33:11 <crc> I440r: how are the registers used in isForth?
14:37:12 <arke> eax = W
14:37:15 <arke> ebx = TOS
14:37:23 <arke> ebp = rstack
14:37:29 <arke> esp = data stacl
14:37:40 <arke> esi =  Forth IP
14:37:46 <arke> :)
14:39:11 <crc> Thanks arke
14:47:32 * kc5tja goes to read up on cache kernel designs.
14:58:25 * arke is writing a _printz_inline routine
15:01:10 --- join: mur_ (~mur@uiah.fi) joined #forth
15:07:35 --- join: FlamingRain (~Ecoder@c-24-129-95-254.se.client2.attbi.com) joined #forth
15:13:34 --- quit: mur (Read error: 110 (Connection timed out))
15:14:34 <kc5tja> See, now I'm wondering if I should just design a simple, single-board computer with no real on-board peripherals, but include some kind of generic yet performant I/O bus or something.
15:15:17 <kc5tja> Basically a very, very, very small mainframe type of thing.  No on-board video, no on-board audio -- etc.  These things would be provided through external plug-in cards.
15:15:24 <kc5tja> What are your thoughts on this?
15:15:49 <Topaz> sure, at least, it'd make design far easier, but how would the bus limit stuff?
15:16:22 <crc> kc5tja: I'd like that :-)
15:16:28 <Topaz> (i mean, long bus wires are generally dangerous)
15:16:37 <Topaz> although LART appears to do it successfully, with a rather cute header
15:16:42 --- quit: imaginator ("dentist")
15:16:52 <kc5tja> I'm not sure what you mean by LART.
15:17:02 <kc5tja> If the bus were operated slowly enough, longer wires wouldn't be an issue.
15:17:18 <kc5tja> It's only when you have to deal with high clock-speeds that issues start to arise.
15:17:51 <Topaz> http://www.lart.tudelft.nl/gallery/hand.jpg
15:18:04 <Topaz> ( http://www.lart.tudelft.nl/ )
15:18:50 --- quit: FlamingRain ("Leaving")
15:19:00 <Topaz> (naturally, i want one ;)
15:20:15 <crc> I'm planning to buy one of these: http://gumstix.com/ and build an FPGA-based expansion card for it :-)
15:22:39 <kc5tja> 230 pounds is an awful lot of money for something like that.  That doesn't bode well for what I was thinking of either.  *sigh*
15:22:43 <kc5tja> Scratch that idea, I guess.
15:24:29 <Topaz> i dunno why they're so expensive, i guess short-run assembly is an issue though
15:25:43 <slava> hi guys
15:26:39 <Robert> Hi slava 
15:27:04 <slava> just got home after walking in heavy rain :)
15:28:48 <Robert> Fun
15:29:00 <Robert> No rain here for a while, just sun...and lots of it.
15:29:08 <kc5tja> Topaz: I'm willing to bet you it's because they get their boards fully manufactured from a professional manufacturing shop.
15:29:20 * kc5tja however will be distributing things as a kit.  
15:29:29 <Topaz> yeah, it's a university project
15:29:36 <kc5tja> Or, at least, I would LIKE to distribute it as a kit.
15:29:36 <Topaz> who always seem to have outrageous budgets
15:30:03 <kc5tja> But I can always hand-assemble the parts before shipping if people pay me enough.
15:30:36 * kc5tja is frightened at the prospect of soldering a 208-pin, surface-mount chip to a PCB.
15:31:00 <Topaz> if you lacquer the board (and get the desoldering wick at the ready) it shouldn't be too difficult, maybe ;)
15:32:54 <kc5tja> Actually, the method I've been told works great for soldering these kinds of chips is to actually form a solder bridge across ALL the pins on one side first, then use desoldering wick to suck the excess off.
15:33:15 <arke> hehe
15:33:20 --- join: mur (~mur@uiah.fi) joined #forth
15:33:36 * arke now playing: Red Hot Chili Peppers - Sexy Mexican Maid
15:33:40 <arke> great song
15:33:41 <arke> :)
15:34:32 * arke gets back to printz_inline
15:45:30 --- quit: mur_ (Read error: 110 (Connection timed out))
15:46:58 <kc5tja> Hehe -- hey, that MIPS processor can be attained at a 400MHz implementation for only $28.  :)
15:47:25 <warpzero> dang
15:47:30 <warpzero> i could make a supercomputer
15:47:51 <warpzero> can you make fpgas to connect cpus that fast
15:48:22 <kc5tja> warpzero: Well, maybe, but at $100 for a low-end FPGA that's been around for 5 years, I wouldn't want to try.
15:49:34 <kc5tja> Although, come to think of it, a $25 MIPS processor and $100 FPGA implementing a lot of the peripheral components pretty much comes close to the $130 maximum price for the Kestrel.
15:49:54 <kc5tja> Though, to make sure I get a good return on investment, I would have to charge at least 50% profit.
15:50:13 <kc5tja> So that'd be close to $220 for the rest of the world.
15:50:20 <kc5tja> And that doesn't include SDRAM.
15:50:41 <kc5tja> Or the rest of the hardware needed to make it work (e.g., like a $20 VGA connector -- man, what a freaking rip!)
15:51:13 <warpzero> you can buy cards cheaper than that
15:51:15 <warpzero> lol
15:51:29 <kc5tja> Yes.
15:52:11 <warpzero> well dude
15:52:20 <warpzero> im off
15:52:22 --- quit: warpzero ("Tried to warn you about Chino and Daddy Gee, but I can't seem to get to you through the U.S. Mail.")
15:59:21 --- quit: crc ("Time for bed... Goodnight all!")
16:27:03 --- join: TheBlueWizard (TheBlueWiz@pc88dn1d.ppp.FCC.NET) joined #forth
16:27:03 --- mode: ChanServ set +o TheBlueWizard
16:31:04 <Robert> Hi, Wizard
16:31:53 <TheBlueWizard> Robert hiya
16:32:05 <Robert> :)
16:58:15 --- join: mur_ (~mur@smtp.uiah.fi) joined #forth
17:08:50 --- quit: mur (Read error: 110 (Connection timed out))
17:18:33 <arke> hi
17:18:37 * arke sweat
17:24:43 --- quit: Topaz (Remote closed the connection)
17:26:20 --- join: mur (~mur@smtp.uiah.fi) joined #forth
17:27:02 <TheBlueWizard> arke hiya...what's the "sweat" thing?
17:27:32 <arke> just came home
17:27:34 <arke> rode bike
17:27:37 <arke> and i am sweating :)
17:27:47 <arke> but other than that, I am coding on F2 :)
17:28:10 <arke> (got a good enough microkernel done, so I'm doing some final tests and then starting on the Forth)
17:28:54 <TheBlueWizard> ah
17:29:08 <arke> but first, I need some water :)
17:38:51 --- quit: mur_ (Read error: 110 (Connection timed out))
17:54:08 <arke> wheee
17:54:16 <arke> i just wrote a TINY _strlen
17:54:39 <arke> actually, I'm not quite done yet.
17:54:47 <arke> i think I can eliminate the use of eax too
17:54:48 <arke> :)
17:54:51 <Robert> Using counted strings? 
17:54:55 <arke> ...maybe
17:54:58 <arke> noope
17:55:03 <arke> stringz
17:55:15 <arke> need the strlen for _printz_inline
17:55:27 <Robert> Hrm.
17:56:09 <arke> nope, i need eax.
17:56:15 <arke> which isn't a problem.
17:57:26 <arke> push eax
17:57:29 <arke> push esi
17:57:33 <arke> .nxt: lodsb
17:57:37 <arke> jnz .done
17:57:41 <arke> err
17:57:45 <arke> jnz .nxt
17:57:51 <arke> pop ecx
17:57:54 <arke> sub ecx, esi
17:57:58 <arke> neg ecx
17:57:59 <arke> pop eax
17:58:00 <arke> ret
17:58:00 <arke> :/
17:58:05 <arke> what do you think?
17:58:42 <arke> rgh, this doens't do what i want it to :/
17:58:51 <arke> hold on
17:59:26 <Robert> There are some fast implementations.
17:59:32 <arke> aah, got it.
17:59:45 <arke> pop ecx
17:59:49 <arke> xchg ecx, esi
17:59:56 <arke> sub ecx, esi
17:59:58 <arke> pop eax
17:59:58 <arke> ret
17:59:59 <arke> :)
18:00:11 <arke> better? :)
18:00:21 <slava> arke, you're writing a forth?
18:04:08 <TheBlueWizard> gotta go....all bye
18:04:18 --- part: TheBlueWizard left #forth
18:12:31 <arke> slava: yes :)
18:12:36 <arke> bye tbw...
18:12:41 <slava> cool
18:12:58 <slava> arke, i added a cool feature to factor. on startup, it calls 'see' with a random word under the banner 'word of the day' :-) :-)
18:13:10 <Robert> Heh
18:13:35 <arke> :)
18:14:20 <arke> ooh!
18:14:34 <arke> i think i mighthave an even better way for the strlne
18:14:54 <slava> Robert, there's 927 words :)
18:15:05 <slava> in 40 vocabularies
18:15:07 <Robert> What a bloat :)
18:15:12 <arke> BLOATWARE
18:15:13 <slava> bah
18:15:13 <arke> :)
18:15:16 <slava> this includes an http server
18:15:29 <Robert> Hmm... I might forgive you then.
18:15:31 <slava> and most of the runtime is not written in C
18:15:35 <Robert> If the words are short.
18:15:39 <slava> of course
18:16:09 <arke> 40 vocabs though? :/
18:16:17 <slava> arke, it keeps things modular
18:16:21 <slava> i don't like one huge flat list
18:16:29 <slava> a vocab for string words, list words, vector words, math, etc
18:16:46 <arke> okies
18:16:47 <arke> :)
18:17:15 <arke> Robert: es is set upon entyry into pmode, right?
18:17:19 * arke is making sure
18:17:27 * arke remembers seeing it though
18:17:58 <Robert> I think so.
18:17:59 <slava> arke, this is closer to a scheme implementation in the scope of the standard library. there's a bit more than your typical forth @ ! + - * / :)
18:18:14 <arke> ^_%^_^__^^_
18:20:05 <arke> nope
18:20:12 <arke> waht i have right now is the best strlen, it seems
18:20:13 <arke> :)
18:20:25 * slava 's strlen looks up the length field
18:20:38 <arke> well, not here.
18:20:50 <slava> C strings? :( :(
18:20:54 <arke> strlen will actually be renamed __count
18:20:56 <arke> yeah 
18:21:38 <arke> except that __count uses the forth stack instead of registers. But that will come later.
18:22:09 <arke> Right now, I am doing some microkernel tests 
18:23:29 <arke> bah.
18:23:33 <arke> i dont need that complexity. 
18:23:36 * arke dd's
18:38:45 <madgarden> kill me if you must, but I found ROLL quite useful today.
18:39:01 <slava> i haven't used roll for months
18:39:04 <slava> i don't have it anymore
18:39:10 <slava> or pick
18:39:13 <slava> i use pick to mean another thing :)
18:39:51 <madgarden> I used it to take the contents of the stack and make a string out of it.
18:40:02 <madgarden> I'll send this string to a remote test client for evaluation.
18:40:03 <slava> .s?
18:40:20 * arke kills madgarden 
18:40:34 <madgarden> Well no, my .s is a bit more specialized.
18:40:37 <slava> madgarden, the string itself is on the stack?
18:40:43 <slava> : .s datastack . ;
18:40:56 <madgarden> yea, the resulting string is on the stack.
18:41:01 <slava> eew
18:41:01 <slava> why?
18:41:15 <madgarden> On the stack I have something like:
18:41:15 <madgarden> 1 2 3 4 "foo" 5 "bar"
18:41:27 <madgarden> And I make a string: " 1 2 3 4 "foo" 5 "bar" "
18:42:05 <madgarden> Why? Because the test client is going to be executing forth on the test server remotely, and wants to get the results back somehow. So, I ask for the stack, get a string, evaluate, and now I have the stack.
18:42:17 <slava> oh
18:42:34 <slava> in factor, if you evaluate the output of .s and  execute set-datastack, it restores the stack :)
18:42:41 <slava> because .s prints { 1 2 3 "hello" } etc.
18:42:46 <slava> and { } is a vector
18:42:49 <madgarden> What if you have a ' word on there?
18:43:54 <slava> i don't have '
18:44:49 <slava> to pass code without executing, i push a list [ 2 3 + . ]
18:44:56 <slava> then execute the list with call ( list -- )
18:45:00 <madgarden> I can have all sorts of crap on the stack. In this case though, I only care about the basic types (int, float, string).
18:45:11 <madgarden> Yea, that's neat.
18:45:17 <slava> all my types except file handles can be turned back and forth into source
18:45:23 <slava> source that recreates that object
18:45:34 <madgarden> That's nutty.
18:45:34 --- join: kc5tja_ (~kc5tja@66.74.218.202) joined #forth
18:47:36 <madgarden> Can you get the XT of an existing word?
18:47:51 <kc5tja_> madgarden: In Forth?  '
18:47:54 <slava> you can get a list containing the definition of a colon def
18:48:13 <madgarden> kc5tja_, I know that silly-head. :P
18:48:15 <slava> you can pass a word itself on the stack
18:48:16 <kc5tja_> Oops, I must have missed some conversation while I was napping.  I didn't realize I got disconnected.
18:48:21 <slava> there is no XT
18:48:25 <slava> and no '
18:48:35 <slava> just call ( list -- ) and execute ( word -- )
18:48:53 <madgarden> Interesting.
18:49:32 <slava> intern ( string -- word ) gets you a word on the stack
18:51:14 <madgarden> kc5tja, I was just describing how I used ROLL today as the simplest solution to a problem.
18:51:30 <madgarden> (which I'm sure you guys could solve in many more clever ways)
18:53:41 --- quit: kc5tja (Nick collision from services.)
18:53:44 --- nick: kc5tja_ -> kc5tja
18:53:45 <arke> EWWWWW ROLL HERESY!!!!
18:53:47 <arke> :P
18:53:51 --- mode: ChanServ set +o kc5tja
18:53:55 * arke goes out to eat dinner
18:54:23 <kc5tja> What application is that?
18:54:26 <madgarden> : r0llz0rzez!!1 depth roll ;
18:54:32 <madgarden> Heh, not that.
18:54:48 <madgarden> THis is my remote test client/server application.
18:55:02 <madgarden> For an embedded platform.
18:55:23 <madgarden> The client will script the test, and execute code on the server. The server will send results back to the client.
18:55:53 <madgarden> So, I figured the simplest protocol would be to get the server to give me its stack when I want it, and then just evaluate that during my tests.
18:56:47 <madgarden> And using the lovely word-overriding features of Forth, I could run the same script as client/server, or standalone on the server if I want.
19:07:19 <madgarden> 01,9910Disgusting?
19:11:23 <slava> madgarden, my httpd server works in cfactor now .
19:11:39 <madgarden> Cool! Is it up?
19:11:42 <slava> not yet
19:11:47 <slava> let me get some of the applications running in it first
19:16:59 --- quit: I440r ("Leaving")
19:17:26 --- quit: [Forth] ("abort" Reality Strikes Again"")
19:54:14 --- quit: tgunr (Excess Flood)
19:56:11 --- join: tgunr (~davec@vsat-148-63-4-107.c001.g4.mrt.starband.net) joined #forth
20:16:48 --- nick: TomasuAway -> Tomasu
20:22:11 --- join: theFox (~jdfox572@adsl-69-104-246-9.dsl.pltn13.pacbell.net) joined #forth
20:23:13 <slava> hi theFox
20:24:11 <kc5tja> re theFox
20:28:23 --- join: imaginator (~George@georgeps.dsl.xmission.com) joined #forth
20:28:47 <kc5tja> theFox: Long time no see.  How are things?
20:30:00 <slava> hi imaginator
20:34:07 <imaginator> Hi slava and others.
20:34:19 <imaginator> Is theFox Jeff Fox?
20:34:40 <kc5tja> imaginator: Yes, but he seems to be idling at the moment.
20:34:47 <imaginator> cool
20:36:39 <madgarden> kc5tja, how does my stack->string remote test app sound... decent way to do things?
20:37:00 <kc5tja> madgarden: I have no idea.  I know absolutley nothing about it.
20:39:44 <madgarden> Well the basic idea is this... 2 forth systems exist, one on the test client, one the server (terminal). The client sends commands to the server... stuff like "getEepromWord" etc. Rather than have the server validate the results, the client will request the results as a string representation of the stack, and then evaluate and validate.
20:42:17 <kc5tja> I would expect the terminal to be the client.
20:43:01 <slava> http://69.198.144.227:8888/inspect/
20:43:20 <kc5tja> But otherwise, I would need to know more about what it is you're trying to accomplish before I can think of things to say about it.  For example, it sounds like the only time the server talks back to the client is when it TYPEs something to the console.  So getting its current stack state can be accomplished with a .S .
20:43:45 <madgarden> Depends how you look at it. We're using the term "client" for the testing PC, anyway.
20:44:07 <slava> madgarden, i have the native http server running now :)
20:44:32 <slava> madgarden, if you want, you can go to http://69.198.144.227:8888/test/ and it will dump the entire test suite output in your browser :)
20:44:47 <madgarden> And, true, that... but... I may not want it to type anything back to the client unless I ask for it. This thing will also double as a "debug shell" of sorts.
20:48:13 <madgarden> slava,
20:48:13 <madgarden> Factor Test Suite
20:48:13 <madgarden> Running Factor test suite...
20:48:13 <madgarden> Testing crashes...
20:48:17 <madgarden> :O
20:48:40 <slava> madgarden, oh, i forgot to upload the test suite sources to my server
20:54:25 <ianp> but scary :)
20:54:28 <ianp> woops
20:54:28 <ianp> hi
21:00:53 --- quit: theFox (Read error: 110 (Connection timed out))
21:24:50 <arke> oh, jeff fox was here :/
21:24:54 <arke> and i missed it
21:24:55 <arke> DAMN
21:25:01 <kc5tja> You didn't miss much.
21:25:16 <arke> yep, seems so.
21:25:20 <arke> kc5tja: do you prefer nasm or gas?
21:25:22 * slava is coding simpson's method in factor.
21:25:34 * arke is considering switching over to gas
21:26:08 <kc5tja> I personally prefer nasm.  But I think if I were to (re)start development on Dolphin again, I'd use fasm.
21:26:25 <arke> fasm is better than nasm.
21:27:15 <arke> but i was thinking if I should switch over to gas because of its increased availability and just the fact that its much clearer than nasm if you know the gas syntax
21:27:18 <arke> or even just partuially
21:28:12 <kc5tja> As long as you are aware that gas is not intended for humans to use, then fine.
21:28:59 <arke> heh
21:29:10 <arke> many use ggas anyway :)
21:29:17 <arke> kernel, all inline asm with gcc
21:30:05 <imaginator> gas is nice.
21:30:13 * imaginator likes the AT&T syntax
21:30:17 <fridge> jumpin jack flash its a gas gas gas
21:30:22 <imaginator> And it has .macro :)
21:30:25 <arke> :)
21:30:27 <arke> :)
21:31:01 <imaginator> The assembler matters less than what you create.
21:35:39 <arke> imaginator: well, i have a question.
21:36:02 <arke> imaginator: is it possible to have a numeric constant be defined, and have a macro which, every time it is called, changes the constant?
21:36:13 <arke> imaginator: its possible in fasm, i don't know about nasm or gas thjough
21:36:17 <arke> probably not in nasm
21:37:50 <imaginator> I've never tried that.
21:38:13 <arke> it makes it much easier to do dictionary upkeep.
21:38:13 <kc5tja> Then, by definition, the constant isn't.  :)
21:38:27 <arke> hrm? oh
21:38:27 <arke> lol
21:38:29 <arke> yeah :)
21:38:47 <arke> kc5tja: won't be able to meet - paretns decided to go to hurst castle instead :/
21:38:59 <arke> or however you spell that :)
21:39:48 <arke> :/
21:41:13 <arke> damn. "If you .set a global symbol, the value stored in the object file is the last value stored into it."
21:42:58 <kc5tja> arke: That's fine.  Whatever -- I'm sure there will be other opportunities.
21:43:05 * kc5tja is eyeballing the L4 microkernel right now.
21:44:00 <arke> heard about that. 
21:48:45 <arke> hmm wait
21:48:59 <arke> argh
21:49:00 <arke> doesn't work.
21:49:01 <arke> :/
21:49:06 <arke> oh well.
21:49:09 <arke> I'll just hack around it.
21:49:10 <arke> :)
21:49:23 <imaginator> arke: you could generate machine code at runtime you know.
21:49:36 <imaginator> It's pretty easy to do.  I found it wasn't as difficult as some may lead you to believe.
21:49:40 <arke> yes.
21:49:55 <arke> its not too hard to do without.
21:50:03 <arke> just might be a bit of a pain, but it wont be bad.
21:50:10 * arke is excited
21:55:46 <kc5tja> Why so excited?
21:55:55 * kc5tja can't remember being excited about anything in life in a long time.
22:00:49 --- join: mur_ (~mur@uiah.fi) joined #forth
22:07:57 <arke> F2 :)
22:08:04 <arke> recoding it in gas
22:08:08 <arke> because then I
22:08:14 <arke> because then I'll get more of a feel for the code
22:08:21 <arke> right now, I just kinda went over it
22:12:19 --- quit: mur (Read error: 110 (Connection timed out))
22:25:23 --- quit: Fractal (Read error: 110 (Connection timed out))
22:38:17 --- quit: mur_ (Remote closed the connection)
22:38:57 --- join: mur (~mur@uiah.fi) joined #forth
22:42:15 <kc5tja> Well, I can now say that L4 may be a nice microkernel, but its API is utterly rancid.
22:42:37 <kc5tja> If folks thought that realloc() was bad, oh my god, you ain't seen nothing yet.
22:42:59 <arke> yep
22:43:05 <arke> i took a look at it for like 2 minutes
22:43:23 <arke> ugly ^_^
22:43:33 <kc5tja> The IPC "function" is pathetically ugly.
22:43:38 <imaginator> kc5tja: so it's not going to takeover the world or the HURD anytime soon?
22:47:50 <kc5tja> I'm sure they employ an interface like that purely for performance reasons -- the more aggregatable functionality you can pack into a single OS call, the lower the amortized cost of the system call is.
22:48:00 <kc5tja> And L4 is known to just totally blow the doors off of Mach.
22:48:02 <kc5tja> Including GNU/Mach.
22:49:16 <kc5tja> The thing is, using L4's IPC mechanisms seems very difficult at first glance, because it does so much, and IPC-related parameters are distributed through a variety of bizarre data structures, where bits have multiple meanings in different contexts, etc.
23:02:27 <kc5tja> At least L4 is Multiboot compliant -- GRUB can be used for a bootloader.  Very convenient.
23:05:36 --- quit: arke (Read error: 110 (Connection timed out))
23:08:54 --- join: arke (~chris@wbar8.lax1-4-11-100-108.dsl-verizon.net) joined #forth
23:12:11 <arke> wtf
23:12:13 <arke> i wont stop bleeding
23:18:22 <kc5tja> Maybe I'll work on my cache kernel design, and release it as L5.  :D
23:18:33 <kc5tja> Oh well.
23:18:34 <arke> wtf
23:18:37 <arke> im still bleediong
23:18:39 <kc5tja> I can see that I totally wasted this weekend.
23:18:54 <kc5tja> arke: Nose?
23:19:01 <arke> no
23:19:03 <arke> leg
23:20:04 <lyca> arke - go ahead.. spill what happened.
23:20:31 <arke> tjust cut myself
23:20:37 <arke> and iripped it open again a while agio
23:20:41 <arke> and wont stop bleedint
23:20:55 <lyca> are you going to live through it?
23:21:17 <arke> if i stiopbleegin
23:21:28 <arke> :)
23:21:34 <lyca> i can tell you are feeling faint by the way you type.
23:21:53 <arke> im not
23:22:04 <arke> im just tryinf to type exceptionally fast because im palyingf a game
23:22:29 <lyca> a realtime game?
23:22:32 <kc5tja> Games....maybe that's what I need to help cheer me up right about now.
23:23:01 <arke> :)
23:23:07 <lyca> kc5tja - want to play a round of spellcast?
23:26:30 <arke> isnt that a card game or something?
23:26:51 <kc5tja> lyca: I don't have that game.
23:27:20 <kc5tja> I have Missile Command on my Atari 2600 emulator though.  :)
23:28:12 <lyca> kc5tja - http://rm-f.net/~james/spellcast
23:28:41 <kc5tja> Cute domain name.
23:28:57 <arke> :)
23:29:07 <lyca> kc5tja - its ornge's domain.
23:33:36 <kc5tja> It doesn't seem to be working for me.
23:34:40 <lyca> kc5ta- oh. what's the problem?
23:35:56 <lyca> kc5tja - are any of these you?  "Newbie/I'mStillLearning" ?
23:47:44 <kc5tja> Newbie was; I closed the browser though.  Nothing was appearing.  And I couldn't join an existing game either.
23:47:59 <kc5tja> Hey; I'm going for some food.  I'll be back in a little bit (circa 30 to 45 minutes)
23:48:35 <arke> ok
23:50:42 --- join: mur_ (~mur@mgw2.uiah.fi) joined #forth
23:51:24 --- join: Fractal (jah@selling.kernels.to.linus.torvalds.at.hcsw.org) joined #forth
23:59:59 --- log: ended forth/04.08.10