| wolfspraul | kristianpaul: it will be difficult to practically help him | 00:05 |
|---|---|---|
| wolfspraul | does he speak chinese? how long is he here? does he just want to buy the slx9 or other things as well? | 00:05 |
| wolfspraul | can he navigate in beijing by himself? | 00:06 |
| wolfspraul | beijing is huge an inefficient, if you are new basically every little thing you want to do outside takes you a full day. after some time (even for Chinese), you can bring it down to "a few hours" | 00:06 |
| wolfspraul | most people will try to stay where they are and let courier services deliver stuff to them :-) but for that you need fluent Chinese to direct people on the phone about what to do... | 00:07 |
| wolfspraul | I can try to help but also a little hesitating to spend a whole day traveling around to help a foreigner buy a 10 USD chip :-) | 00:07 |
| kristianpaul | nope chinesee.. not that i know | 00:11 |
| kristianpaul | he ask me if i need something | 00:11 |
| kristianpaul | i said slx9 ;) | 00:12 |
| kristianpaul | to be honest i dont bet he would do it, but i have to give him the information | 00:12 |
| kristianpaul | dont help i'm betting him :D | 00:13 |
| kristianpaul | he will be another days i think 4 | 00:13 |
| kristianpaul | and i guess want buy other stuff out of my concern | 00:13 |
| kristianpaul | huge an inefficient, :-/ | 00:14 |
| kristianpaul | wolfspraul: no please dont help ;) | 00:14 |
| wolfspraul | he can buy the slx9 off-the-shelf in a little booth not more than 100m from a subway exit | 00:14 |
| wpwrak | huge and inefficient, like any well-run bureaucracy :) | 00:15 |
| kristianpaul | wpwrak: ;) | 00:15 |
| wolfspraul | BUT... which booth? which subway station and exit. and still need to make it through a gazillion of other confusing signs booths people etc. | 00:15 |
| kristianpaul | hehe | 00:15 |
| wolfspraul | there must be hundreds of places selling those in beijing, but tracking one down specifically requires some knowledge of locality. in china you would just use qq or phone/instant messaging, then track down the source or have them send to you (much easier) | 00:16 |
| kristianpaul | he is bit far looking at the map for chanpeig, beijing.. | 00:16 |
| wolfspraul | as I said - everything is far :-) | 00:16 |
| wolfspraul | not just he | 00:16 |
| kristianpaul | he traveld to visit some indutrial parks | 00:16 |
| kristianpaul | jaja | 00:16 |
| kristianpaul | parks may not be right word.. | 00:16 |
| kristianpaul | qq :) | 00:17 |
| wolfspraul | maybe inefficient is the wrong word. most people would have stuff delivered to them | 00:17 |
| kristianpaul | xiangfu gave some some data perhaps he could call it if know chinesse.. | 00:17 |
| wolfspraul | then you don't care how inefficient it is, because some poor guy will freeze off his fingers delivering the goods to you for 1-2 USD | 00:17 |
| kristianpaul | call the shop! | 00:17 |
| kristianpaul | not xianfgu =) | 00:18 |
| wolfspraul | yes exactly | 00:18 |
| wolfspraul | call, order, deliver | 00:18 |
| kristianpaul | exactly | 00:18 |
| wolfspraul | but - how to pay? how to talk on the phone? | 00:18 |
| kristianpaul | he | 00:18 |
| wpwrak | ah, it's winter over there :) | 00:18 |
| wolfspraul | chinese can pay from their phone in seconds, but a foreigner without cards/accounts/banks/etc will have trouble with those things | 00:18 |
| kristianpaul | ah good point | 00:18 |
| kristianpaul | hum hum | 00:18 |
| kristianpaul | how do you pay? | 00:18 |
| kristianpaul | phone? ;D | 00:18 |
| kristianpaul | i dont bet :) | 00:18 |
| wolfspraul | I keep myself out of those things | 00:19 |
| kristianpaul | anyway he just wanted to advertise me his recent travel | 00:19 |
| wolfspraul | but ok if I would have to I would set it up, as it's the only way to get things done fast | 00:20 |
| kristianpaul | so i ask for that.. lets see how it results | 00:20 |
| kristianpaul | ha if i could call and speak chinesse ;) | 00:21 |
| kristianpaul | anyway, hi ! | 00:21 |
| kristianpaul | did you looked and J1 forth cpu btw? | 00:21 |
| wolfspraul | yes sure | 00:22 |
| wolfspraul | that was a great pdf link | 00:22 |
| kristianpaul | if you add clock support and bram basically thats the shortest path to a working cpu i guess you tought that time ago :) | 00:22 |
| kristianpaul | great as in... | 00:22 |
| wolfspraul | 'look' may be too much though, I just skimp over it and mentally file somewhere | 00:22 |
| kristianpaul | good reading or soemthing else | 00:22 |
| kristianpaul | ? | 00:22 |
| wolfspraul | well one by one | 00:22 |
| wolfspraul | I'm working towards the blinking led, right? | 00:22 |
| kristianpaul | yes | 00:22 |
| kristianpaul | shift register i remember? | 00:22 |
| kristianpaul | this version | 00:22 |
| kristianpaul | next* | 00:22 |
| wolfspraul | blinking led is not using a shift register | 00:23 |
| wolfspraul | but yes, those are missing too, same as tons of other things | 00:23 |
| wolfspraul | I had to do some more groundwork in recent weeks | 00:23 |
| wolfspraul | too bad, but ok, hopefully it will help in the long run | 00:23 |
| kristianpaul | but i will? (shift reg or any other eq counter implementation? ) | 00:24 |
| wolfspraul | you mean whether I will support it? | 00:24 |
| wolfspraul | of course, absolutely | 00:24 |
| wolfspraul | I want to support all of the chip's features | 00:24 |
| wolfspraul | and shift registers are quite high on the list | 00:24 |
| kristianpaul | ahmm | 00:24 |
| wolfspraul | but first the basic luts and clocks | 00:24 |
| wolfspraul | then bram, shift regs, jtag, spi, macc | 00:24 |
| kristianpaul | ok | 00:26 |
| wolfspraul | kristianpaul: what do you like about the j1 cpu? | 00:30 |
| wolfspraul | any specific plans? | 00:30 |
| kristianpaul | learn forth ;) | 00:32 |
| kristianpaul | no more plans | 00:32 |
| kristianpaul | i like things with single porpused | 00:32 |
| kristianpaul | well i used to that when programing microcontrollers... | 00:32 |
| kristianpaul | i like the most i dont need a fancy compiler | 00:33 |
| kristianpaul | is small and simpler | 00:33 |
| kristianpaul | and the closest thing i see you can implement with fpgatools :) | 00:34 |
| wolfspraul | ok but sounds like you are mostly wishing those things | 00:34 |
| wolfspraul | how can we bring it into reality? | 00:34 |
| kristianpaul | the fpgatools part yes | 00:34 |
| wolfspraul | one thing about forth seems to be that they are quite good at articulating something people may want | 00:34 |
| kristianpaul | the learning part i could, but yes for now is just a wishlit | 00:34 |
| wolfspraul | but where is the actual realization? | 00:34 |
| wolfspraul | "don't need a fancy compiler" | 00:34 |
| wolfspraul | sounds great | 00:35 |
| kristianpaul | haha | 00:35 |
| wolfspraul | but show me something that actually runs | 00:35 |
| kristianpaul | :-) | 00:35 |
| wolfspraul | maybe then it doesn't sound so great anymore that "you don't need a fancy compiler"? | 00:35 |
| wolfspraul | yes so that all reads well to me, but I want to make it run and do something - not just some fancy words | 00:35 |
| kristianpaul | i know what you mean... | 00:36 |
| wolfspraul | I want a computer that reads my mind and does what I want it to do | 00:36 |
| wolfspraul | then I only need one command "do" | 00:37 |
| wolfspraul | wouldn't that be cool? | 00:37 |
| kristianpaul | jaja | 00:37 |
| wolfspraul | I should write a paper about it... | 00:37 |
| wolfspraul | do! | 00:37 |
| Last message repeated 1 time(s). | 00:37 | |
| wolfspraul | cool | 00:37 |
| wolfspraul | it's also perfect for kids | 00:37 |
| wolfspraul | oh wait, kids mostly say "give me" or "I want" | 00:37 |
| wolfspraul | hmm | 00:37 |
| kristianpaul | okay to much coffe :) got it! | 00:37 |
| kristianpaul | lol | 00:38 |
| wolfspraul | nah I like the paper, but I want to know how to make it work | 00:38 |
| wolfspraul | so let's make it real, and if nothing real I will dump it :-) | 00:38 |
| wolfspraul | coffee yes, back to fpga switches... :-) | 00:39 |
| LunaVorax | Hi | 09:46 |
| wpwrak | writing compilers isn't all THAT hard. plus, who cares if it's fancy if it runs on a PC ? | 12:51 |
| wpwrak | (writing compilers) that is, for a reasonably simple language. of course, if you want to beat gcc or llvm, there'll be some work involved | 12:54 |
| larsc | writing _good_ compilers on the other hand is a different story | 12:54 |
| wpwrak | yeah. but a simple one that lets you write code that blinks a led while a button is pressed or such things, and that doesn't have to be particularly efficient, would be quite simple. | 12:57 |
| larsc | I'm currently taking the coursera compiler class, it's quite good imo | 12:58 |
| larsc | https://www.coursera.org/course/compilers | 12:59 |
| larsc | need to write the semantic checker tonight | 12:59 |
| wpwrak | wirth's book on compiler design covers one such language quite nicely. i read the pascal-ish one. this seems to be an updated version: http://www.google.com/url?q=http://www.ethoberon.ethz.ch/WirthPubl/CBEAll.pdf&sa=U&ei=4HKvUKPRM46F0QHuuICAAg&ved=0CBUQFjAA&usg=AFQjCNGxDramddD3bwOYKYZmBKDggAp34A | 12:59 |
| wpwrak | argh | 12:59 |
| wpwrak | www.ethoberon.ethz.ch/WirthPubl/CBEAll.pdf | 12:59 |
| Action: wpwrak hates google's uncopyable search results | 13:00 | |
| wpwrak | with 131 pages it's a refreshing difference from the other tomes usually sized like epic fantasy novels | 13:01 |
| wpwrak | oh, and with yacc at your disposition, you get operator precedence basically for free. yet another one of the scary bits someone else is taking care of for you. | 13:02 |
| wpwrak | i think the old one is "compilerbau". 118 pages :) | 13:06 |
| viric | wpwrak: I also hate them (search results) | 13:22 |
| viric | with a pdf, it gets specially annoying | 13:22 |
| viric | but I think they overcome that with javascript tricks, if you had javascript enabled :) | 13:22 |
| wpwrak | javascript is on, of course. there's very little that works without it :-( | 13:32 |
| GNUtoo-desktop | wpwrak, maybe use duck duck go then? | 13:32 |
| wpwrak | that's an option, yes | 13:33 |
| viric | ah ok | 13:33 |
| viric | duckduckgo uses bing | 13:33 |
| viric | iirc | 13:33 |
| GNUtoo-desktop | it uses many things | 13:34 |
| viric | use seeks | 13:34 |
| GNUtoo-desktop | it probably uses a combinaison of bing, google, yahoo and its own stuff | 13:34 |
| viric | I'd say it uses bing, and only some times, it gives a special big link of its own cook | 13:35 |
| Action: GNUtoo-desktop doesn't understand why what backend it uses is important | 13:36 | |
| viric | I thought wpwrak meant *google* | 13:38 |
| viric | seeks will use google. | 13:39 |
| mth | wpwrak: how does that work? (yacc doing operator precendence for you) | 15:09 |
| mth | I've always coded compilers by hand, but operator precendence is determined by how you write your grammar, not by how you code it | 15:09 |
| mth | I thought yacc takes something akin to EBNF as input; if so the operator precedence should be already determined in the input | 15:10 |
| larsc | you tell yacc the precendence order and it will take care of it | 15:10 |
| mth | it has a more high-level input format that? | 15:10 |
| mth | s/that/then/ | 15:11 |
| wpwrak | larsc: that's the dirty way | 15:11 |
| wpwrak | mth: and yes, what i mean is the grammar | 15:11 |
| larsc | well you use the precendence order to resolve ambiguities in your grammar | 15:12 |
| wpwrak | larsc: you mean in your sloppy grammar :) | 15:12 |
| larsc | keeps the grammar much simpler | 15:13 |
| wpwrak | mth: e.g., a recursive-descent parser can't do proper operator precedence (without adding extra twists) | 15:13 |
| mth | you only need a look ahead of one symbol to be able to do it | 15:14 |
| wpwrak | larsc: you have to think a little harder. but that's often a good thing :) | 15:14 |
| larsc | wpwrak: I need to do the transformations that yacc does for me by hand | 15:14 |
| larsc | mth: the grammar also has to have other certain properties | 15:15 |
| wpwrak | larsc: and by doing this, you purge your grammar from "magic". makes it easier to resolve conflicts you'll run into later. | 15:16 |
| larsc | I belive in magic ;) | 15:16 |
| mth | larsc: yes, but it is implementable with a recursive descent parser with a lookahead of just one, if your grammar satisfies those properties | 15:17 |
| larsc | believe | 15:17 |
| wpwrak | that one's the bad kind :) | 15:17 |
| mth | magic is great if it works, but as soon as it breaks down you'll be wishing you'd be dealing with something simpler | 15:18 |
| larsc | iirc, you use the lookahead to resolve shift-reduce-conflicts | 15:18 |
| larsc | if your grammar does not have any you can also use a parser with 0 lookahead | 15:18 |
| mth | it's been a while since I did compiler theory, but doesn't 0 lookahead put you somewhere on the level of regular expressions? | 15:20 |
| mth | or maybe not even that, just a finite state machine | 15:22 |
| larsc | you have a stack don't you? | 15:22 |
| mth | what does lookahead 0 mean anyway? I guess that once you see the symbol you have to immediately consume it (no peek operation) | 15:24 |
| larsc | iirc with lookahead of one you look at the next terminal symbol and based on that decide how to handle your current symbol | 15:31 |
| larsc | with lookahead of zero you don't look at the next terminal, but just use your current symbol | 15:31 |
| mth | by that definition you can even do operator precedence with a lookahead of 0 | 16:10 |
| kristianpaul | wpwrak: you like tiny boards, check this http://keithp.com/blogs/MicroPeak/ | 23:51 |
| wpwrak | they;re wasting a lot of space | 23:54 |
| kristianpaul | lol | 23:54 |
| kristianpaul | check backwards the battery socket | 23:55 |
| wpwrak | yup. with the through-hole variant, the board could be smaller | 23:55 |
| wpwrak | by some 4 mm | 23:55 |
| kristianpaul | ahm yes | 23:57 |
| --- Sat Nov 24 2012 | 00:00 | |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!