#avr Logs

Jul 19 2019

12:05 AM nohit: --
01:09 AM Maya-sama is now known as Miyu
05:53 AM djph: ignore -channels #avr * JOINS PARTS QUITS
05:53 AM djph: ..
05:53 AM djph: note to self, get coffee
06:42 AM davor_ is now known as davor
08:51 AM vmt: rue_mohr: stop doing javascript
08:52 AM vmt: it's like weed, it will only serve to dumb you down
08:52 AM polprog: can confirm
08:53 AM polprog: im using jQUery which is like a bong, makes it more pleasant to use (over raw JS)
08:53 AM polprog: and you can put flowers in it when the mother is visiting
08:53 AM vmt: is it really that hard to build a c/++ api in browsers exposing the api and running the applications in a sandbox
08:53 AM vmt: i mean
08:53 AM vmt: exposing the dom.
08:53 AM polprog: yeah. webdev sucks
08:54 AM vmt: first, we need js. then we need some fucking ultra dumbass "webasm"
08:54 AM skz81: polprog, my own mother would ROFL seeing that
08:55 AM polprog: luckily there is CGI
08:55 AM vmt: not on client side
08:55 AM polprog: oh, yeah
08:55 AM polprog: right
08:55 AM vmt: i don't care about server-side. you can do whatever there.
08:55 AM vmt: all you need is to be http compliant. 1.1 is enough
08:56 AM polprog: dunno, ive learnt to live with JS, and every day i dont have to touch it is a happy day
08:56 AM vmt: though i guess you can write js scripts to do cgi, no?
08:57 AM vmt: i don't really know, then again i really don't care either. furthermore, it's still not the same
08:57 AM vmt: and i figure security niggers will be all over the place is someone does that
08:57 AM vmt: if*. fingers aren't working
08:59 AM skz81: security fingers ??
08:59 AM vmt: no :(
08:59 AM * skz81 put down the bong
08:59 AM vmt: i mean can you actually write js which executes arbitrary programs client-side?
09:01 AM skz81: I guess it's theorically feasible, yes... But 1) over-engineered 2) possible security holes as you mentionnnd
09:02 AM skz81: Unsure spawning a process for each request is a good idea anyway
09:04 AM vmt: well no, 1) there's a lot of "single-page applications" infested with 100ks LOC of js bullshittery, 2) technically you could start a process which listens to calls on a socket, the calls would need to have some kind of a security token...
09:05 AM vmt: but even then it's a poor solution because the dom would need to be exposed via somekind of a javashit bridge
09:07 AM skz81: Ho sorry, i misread
09:07 AM skz81: client-side
09:07 AM vmt: yes
09:07 AM vmt: server-side javascript is idiotic at best, in any case
09:07 AM skz81: would be a huge security hole I guess !
09:08 AM skz81: BUT
09:09 AM skz81: If the user as installed a server-like executable on her machine, plus launched it.... The page could make request to it !
09:10 AM skz81: (can be auto-launched I s'pose)
09:10 AM djph: skz81: if they're doing that, then just use C :)
09:11 AM vmt: you mean through local loopback?
09:11 AM vmt: not have the request do a merry-go-round from the server?
09:13 AM vmt: i mean, user loads some js from domain.com, executes js which does a request (ajax?) onto localhost?
09:15 AM vmt: wait i'm lost in topology here... nevermind. anyway, even if it would work it's a piss-poor solution. i advocate for a simple api which exposes the dom and doesn't necessitate javashit or webshitsembly
09:17 AM vmt: run the executable with limited permissions and finally you'd have something reasonable
09:22 AM vmt: i guess this wouldn't fly either because dynamic linking is *the thing to do*, and because of fads like rust where dependencies are pretty much encouraged
09:23 AM vmt: you would need to have some kind of a dep resolution step before runtime... software is such a fucking hot mess
10:07 AM kre10s: hi. Anyone know what the format for the instruction set of the PMS150 is?
10:59 AM vmt: kre10s: rtfm, http://www.padauk.com.tw/upload/doc/PMC150,%20PMS150%20datasheet%20V108_EN_20181211.pdf
12:01 PM Kliment: vmt: padauk makes some really fun parts
12:01 PM vmt: from the looks of it they make some really cheap ones
12:02 PM Kliment: vmt: Yeah, it's the micros used in stuff like toys and musical greeting cards and such
12:02 PM Kliment: vmt: they are very unreliable because 100% memory testing is not doable at that price
12:02 PM vmt: :D
12:03 PM vmt: do you run readback tests then or what's the deal. or just go all \:D/ about it
12:03 PM Kliment: They make no secret of this - they recommend in-system go-nogo functional testing
12:04 PM Kliment: If all your thing needs to do is make warbly noises and blink LEDs then it's easy to check if it does it or not
12:04 PM Kliment: If you need anything more sophisticated, this is probably not the microcontroller for your application
12:04 PM vmt: yeeah. oh it's otp progmem
12:04 PM Kliment: Since you can get fully tested ones for 3x the price
12:04 PM Kliment: And it's still ten cents
12:05 PM Kliment: But of course for toys or disposable stuff every cent counts
12:05 PM vmt: how do you program these things
12:05 PM vmt: is the interface public?
12:05 PM Kliment: vmt: There is a dedicated high voltage programming fixture
12:05 PM Kliment: vmt: They sell it
12:05 PM vmt: "fixture" means some proprietary stuff?
12:05 PM vmt: sure
12:05 PM vmt: how much is that then
12:06 PM vmt: also do you use their assembler or what can you program these things with?
12:07 PM Kliment: vmt: Let me dig out the aricle by someone who used them
12:07 PM vmt: the datasheet specifies the mnemonics but not the actual machine code
12:11 PM Kliment: vmt: check this out https://github.com/cpldcpu/SimPad
12:12 PM Kliment: vmt: it's an open hardware cleanroom reimplementation of the protocol
12:12 PM vmt: sure. apparently sdcc can target somethingsomething...
12:12 PM Kliment: https://www.eevblog.com/forum/blog/eevblog-1144-padauk-programmer-reverse-engineering/
12:24 PM kre10s: vmt: the datasheet contains the instruction set... but not how to write them in binary.
12:24 PM kre10s: how do i go fom mov x, y to some binary word?
12:54 PM djph: kre10s: you mean how do you go from the asm mnemonic to the actual instruction word?
01:03 PM kre10s: yup
01:46 PM vmt: kre10s: https://free-pdk.github.io/PADAUK_FPPA_13_bit_instruction_set.html
01:46 PM vmt: apparently reverse engineered by some ruski
01:47 PM vmt: https://free-pdk.github.io/ there are 14/15/16 bit isas... pms150 is supposed to use 13(??) I don't know. Look up sdcc source maybe.
01:47 PM vmt: moral of the story: buy chinese, get chinese, and get fucked.
01:49 PM timemage: sounds vaguely like how pic operates.
01:58 PM djph: ugh, whoever wrote these documents should be drawn and quartered.
04:39 PM polprog: https://polprog.net/blog/ddd/
05:09 PM vmt: something about the state of l00nix and foss says that the best debugger frontend is from 10+ years ago and looks like it was from 30+ years ago
05:10 PM vmt: then again loonix people love writing shitty python, javascript and rust (rofl) so who needs debuggers riiiight
05:10 PM vmt: just use ctrl-x-a on gdb!11 it's so good!
05:20 PM timemage: vmt, i usually end up typing lay n and i just puts me into tui mode =P
05:22 PM vmt: the tui mode is fucking bullshit
05:22 PM vmt: and so is the regular mode too, whatever that might be called
05:24 PM vmt: i can't imagine anyone enjoying using the tui mode if they aren't wearing stallman-branded shades
05:40 PM polprog: honestly i can usually go about printf debugging
05:40 PM polprog: also, i have no idea why ddd is so ancient
05:40 PM polprog: and noone made anything better
05:40 PM polprog: radare... is barely usable
05:40 PM polprog: it has nice color but its a disassembler after all
05:41 PM LeoNerd: I find printf debugging (or "debug by LEDs" on smaller MCUs) perfectly fine for 90% of the time
05:42 PM polprog: oh yeah. on mcu leds all the way down
05:42 PM polprog: however i once managed to edit a variable inside an AVR with DDD
05:42 PM polprog: that was so silly
05:46 PM polprog: but yeah
05:46 PM polprog: i dont get all the loonix nerds
05:47 PM polprog: maybe im just not crazy enough yet to spend 30 mins setting up breakpoints and watches in gdb
05:47 PM polprog: using an interface which is basically typewriter compatible
05:47 PM polprog: (yeah i know there are some pimp scripts for gdb)
05:47 PM polprog: but i have DDD which is fine for me
05:47 PM polprog: and runs on BSD as well
05:49 PM kre10s: please dont bash on the foss tools. I have to use windows tools at work... and they don't get better.
05:50 PM polprog: the foss tools are like 100x better
05:50 PM polprog: however WinDBG and OllyDBG have several good features which DDD is missing
05:52 PM kre10s: I also "grew up" on led/printf debugging... but when you are given some legacy code breakpoints and the ability to inspect is nice to have.
05:52 PM polprog: yeah. that too
05:53 PM kre10s: problem with debugging embedded stuff is the timeout/events and sleep modes.
07:10 PM soul-d: found simulide and simultron yesterday very limited but was able to dump signals to vcd file with simultron and look at it in gtkwave sadly simultron has almost no components so spent a day building something that sorta pretended to be like hc595 probably have to look at scripting simulations or something
07:11 PM polprog: what are you simulating?
07:12 PM soul-d: my old projects got all my breadboard out the closeth again so downloaded the hex converted to bin > elf and loaded it up in those apps
07:17 PM soul-d: currently trying to figure out some obscure sed1520 lcd i got
07:24 PM soul-d: but switching it different obscure one where i at least have half a datasheet
07:24 PM polprog: so an avr simulator
07:36 PM soul-d: yeah enough played need some sleep now
07:43 PM rue_shop1: ah yes, so you can make programs that only work in simulators
07:45 PM polprog: goodnight
08:02 PM rue_shop1: night
08:08 PM vmt: i was talking about x86 rather than mcus
08:09 PM vmt: where debugging memory especially with custom allocators is slightly trickier than just printfing it
08:10 PM vmt: msvs frontend ported to loonix please.
09:01 PM ferdna: i am trying to get 24v out of the epson pinout... how can i do this? https://i.imgur.com/0qwAUh3.png
09:15 PM rue_shop1: hu?
09:16 PM rue_shop1: technically pin 4 and 6, but ...
09:16 PM rue_shop1: why do printers have strange CD interfaces?
09:30 PM ferdna: rue_shop1, so if i put my oscilloscope in pin4 + and pin6- i would be able to read 24v right?
09:30 PM rue_shop1: volt meter is a better choice
09:31 PM rue_shop1: cause the ground of the scope and the ground of the pin 6 arn't neccissarily the same
09:31 PM rue_shop1: and if their not, something shorts
10:08 PM ferdna: rue_shop1, the multimeter i have is auto ranging... meaning slow... not fast enough to catch a 25ms pulse...
10:16 PM rue_shop1: oh
10:16 PM rue_shop1: I see
10:16 PM rue_shop1: huh
10:42 PM ferdna: rue_shop1, how should i connect the scope?
11:19 PM ferdna: rue_shop1, any ideas?
11:22 PM day__ is now known as day