#avr Logs
Apr 01 2018
#avr Calendar
12:02 AM rue_mohr: aha, found a macro under it
12:26 AM rue_mohr: hmm
12:26 AM nuxil: lolz
12:26 AM rue_mohr: the design of the stm32 io pins are such that flipping a pin between input and output to make it sink or not isn't the same
12:26 AM rue_mohr: it looks like you specifically set them to be open drain with a pullup
12:27 AM rue_mohr: 01: General purpose output Open-drain
12:27 AM rue_mohr: k
12:27 AM nuxil: 'it just sound like a mess
12:28 AM rue_mohr: so I have to set up the bitbang i2c on the stm32 a bit different
12:28 AM rue_mohr: no, it makes sense, its just the other way of looking at it
12:28 AM rue_mohr: instead of flipping it between input and output like an avr, you set it to be a open drain pin, and use the data register to control it
12:29 AM nuxil: hyands
12:29 AM nuxil: *erm
12:30 AM rue_mohr: so, yea, I'll test this on an avr and then play with porting it
12:31 AM nuxil: port the i2c asm code
12:31 AM rue_mohr: I did
12:31 AM rue_mohr: to C
12:31 AM nuxil: :)
12:31 AM rue_mohr: I'll make something that works with flurrys normal i2c lirbary, then I'll try mine
12:31 AM rue_mohr: it should work, the C translation is not as good, but its close
12:32 AM rue_mohr: ok, off to the shop...
12:48 AM rue_shop4: heh, I think I lost of pcf8574
12:51 AM rue_shop4: unless I didn't sort it yet
12:52 AM Ameisen: had a migraine all day
12:52 AM Ameisen: :(
12:52 AM Ameisen: rue_shop4 - some of my low level stuff ends up calling like 4 functions to do it. In the end, it gets compiled into one or two instructions.
12:52 AM Ameisen: Thanks to the beauty of templates.
12:54 AM rue_shop4: C++
12:54 AM rue_shop4: just say no!
12:55 AM Ameisen: Why?
12:55 AM Ameisen: I use C++ for almost everything.
12:55 AM rue_shop4: ok, wtf did I do with all the pcf8574!?!?!
12:55 AM Ameisen: Did you eat them?
12:55 AM rue_shop4: maybe
12:55 AM rue_shop4: they ARE chips...
12:56 AM Ameisen: Still working on a new build tool in C++, and an ARM firmware in C++
12:56 AM Ameisen: maaaay port the firmware to AVR
12:56 AM Ameisen: not sure yet
01:03 AM rue_shop4: ok, this is stupid
01:03 AM rue_shop4: I'd not have lost them
01:03 AM rue_shop4: I wonder how many I'm looking for
01:05 AM rue_shop4: wtf
01:05 AM rue_shop4: according to the history, I never bought any
01:06 AM rue_shop4: that would be consistent with my inability to find any
01:07 AM rue_shop4: I'm sure I have t little tube of 5 or 10 of them
01:07 AM rue_shop4: ...
01:08 AM rue_shop4: ....
01:12 AM rue_shop4: found it
01:12 AM rue_shop4: origionally a tube of 5
01:12 AM rue_shop4: ok, I think they are cheaper than elephant snot, lets fix this supply issue
01:13 AM rue_shop4: oo their not
01:15 AM rue_shop4: $2/5
01:15 AM rue_shop4: ok, the 5 I have is enough
01:20 AM rue_shop3: I can use 74 595/164 for io
02:08 AM rue_shop3: ok, with flurrys library, I had to adjust my pullups to get it to work
02:08 AM rue_shop3: *up* to 10k (!??!??!)
02:33 AM rue_shop3: yaaay!
02:34 AM rue_shop3: ok I ported flurrys asm to C
02:34 AM rue_shop3: not too many errors either
02:37 AM rue_shop3: suppose I should maul it before taking it to stm32, where its gonna get mauled more
02:54 AM rue_mohr: ooo
02:54 AM rue_mohr: interesting, compiling the same source on another machine resulted in the software delay comming out different
03:09 AM nohitzzz: morning
03:12 AM rue_mohr: no
03:15 AM rue_mohr: ok, I think I can port that to stm32
03:59 AM rue_mohr: yes, 32 bits is better
03:59 AM rue_mohr: void Delay(uint32_t d) {
03:59 AM rue_mohr: I no longer need to do two nested 16 bit calls to my delay function
03:59 AM rue_mohr: its faster with 32 bits
04:00 AM rue_mohr: i2c_master.h:16:24: fatal error: avr/io.h: No such file or directory
04:00 AM rue_mohr: #include <avr/io.h>
04:00 AM rue_mohr: my stm32 port FAILS
04:02 AM Thrashbarg: :/
04:03 AM rue_mohr: there, I have compilation!
04:04 AM Thrashbarg: good, now ship it :P
04:04 AM rue_mohr: stm32 might end up with an identity issue tho
04:04 AM rue_mohr: hah
04:05 AM rue_mohr: 8-| apparently I can read a 1/16w thruhole resistor from 3 feet
04:05 AM Thrashbarg: righto
04:06 AM rue_mohr: :( no i2c
04:07 AM rue_mohr: wait, which pin is clk
04:08 AM rue_mohr: oops
04:08 AM rue_mohr: :))))
04:08 AM rue_mohr: YAY FLURRY!
04:08 AM rue_mohr: we have stm32 i2c!!!!
04:08 AM Thrashbarg: cool
04:14 AM rue_mohr: well, its only 2am
04:15 AM rue_mohr: hmm
04:15 AM Thrashbarg: hmm
04:15 AM * rue_mohr looks around the room for soemthing to do
04:15 AM Thrashbarg: Sun 1 Apr 18:43:55 ACST 2018
04:15 AM rue_mohr: I'd post the code to the internet, but it is the 1st of april
04:15 AM Thrashbarg: yeah...
04:15 AM rue_mohr: oooh whats on hackaday!
04:15 AM Thrashbarg: heh
04:16 AM rue_mohr: and whats the geek news site I cant remmeber
04:16 AM Thrashbarg: what slashdot
04:17 AM rue_mohr: yea!
04:18 AM rue_mohr: their apr1 are the best
04:21 AM antto: Ameisen i remember when dealing with SDCC and pics that there were some compiler flags like -smallstack or some such
04:22 AM antto: can the stack size be limited? specifically in gcc/avrgcc?
04:23 AM nohitzzz: so you need that 4K buffer so you can change patterns on the fly ?
04:24 AM nohitzzz: otherwise there is small delay?
04:24 AM antto: 20ms ain't small in my case
04:24 AM antto: the original device from 1982 doesn't have such problem
04:24 AM antto: so it'd be embarassing if mine did
04:47 AM rue_mohr: :) my stm32 is flashing an led
04:47 AM cehteh: gcc has a small int option and some other things can reduce stack usage, but there are iirc no stack checks on AVR, there is some machinery you can compile in, but i never looked at it for AVR's prolly too costly
04:48 AM cehteh: you better do some static analysis to figure out how much stack can be used at any time and/or put some canarys on the end of your ram and check them regulary
04:50 AM rue_mohr: ok, I can think
04:51 AM rue_mohr: I wanted to look at converting this to being interrupt driven
04:51 AM antto: cehteh i don't use "int" i pretty much always use the types from inttypes/stdint and use the smallest one that does the job
04:51 AM rue_mohr: ditto
04:52 AM rue_mohr: since I found out that most of the 'char' types I was using had been turned into 16 bit
04:53 AM cehteh: antto: that wont save you, arithmetic in C is on int by default (esp for smaller types)
04:54 AM antto: i have some const LUTs which i can move into progmem
04:54 AM antto: but i really wish i could know what's happening
04:54 AM cehteh: thats ansi C .. there is a --int8 option (or how its named)
04:55 AM cehteh: but that breaks a lot libs, you basically have to recompile the whole runtime for that
04:55 AM antto: nah, i don't want to go that much crazy
04:56 AM cehteh: i usually dont
04:57 AM cehteh: the compiler can optimize to smaller types in many cases, but not always
04:57 AM antto: what's the penalty for reading a byte from progmem instead of directly from RAM?
04:59 AM cehteh: progmem is word-addressed (16bits)
05:00 AM cehteh: reading single bytes needs some tricks around, more or less optimized
05:01 AM cehteh: but thats all, often the biggiest penalty comes that you have to copy things from progmem to temporary storage (ram, registers) before use
05:01 AM antto: so it's slower
05:02 AM cehteh: depends on the usage
05:02 AM cehteh: when you do some clever tricks it should be the same speed
05:03 AM antto: you see, i ain't too clever ;P~
05:04 AM cehteh: i'd have to look into the details, but having 16bit instead 8bit integers in the LUT may improve it, or interleave 2 LUT's maybe (i havent tried, just guessing here)
05:04 AM antto: oh, i don't think i can use such a hack
05:04 AM cehteh: what lut is that? sometimes some arithmetic deduction instead storing a lut is faster
05:05 AM antto: one of the LUTs is a table of 256 values (8bit) of pre-calculated CRC8 values
05:05 AM cehteh: ah ok
05:07 AM cehteh: i guess there is no *simple* perfect hash for that :D
05:07 AM cehteh: other than crc8 itself
05:07 AM antto: i can't change that because the communication protocol is "set in stone"
05:08 AM cehteh: i meant generating this table arithmetically
05:08 AM cehteh: https://en.wikipedia.org/wiki/Perfect_hash_function
05:09 AM antto: yes, the original code had functions for calculating the CRC8 directly, and a commented-out version of the function that uses this precalculated table
05:09 AM cehteh: you still dont understand :D
05:10 AM antto: :/
05:10 AM cehteh: perfect hash == a function which behaves like a LUT
05:10 AM cehteh: like when you have a table int lut = {1,2,3,4,5,6,7}
05:11 AM cehteh: then you could simplify it by int lut(int index) {return index+1;}
05:11 AM antto: yes, that works on data which doesn't look like WTF
05:12 AM cehteh: perfect hashing generalizes this by funding an arithmetic function which looks like WTF :D
05:12 AM antto: yeah, i get it.. but i'm no math guru
05:14 AM rue_mohr: I used hashes for 4 byte command strings once
05:14 AM rue_mohr: they were 32 bit hashes
05:15 AM rue_mohr: fixed hashes tho 'HELP' 'QUIT', I cant up with lots of random hashes for the 4 byte command strings
05:15 AM rue_mohr: uint32_t QUITHASH = 'QUIT'
05:15 AM rue_mohr: ;
05:16 AM rue_mohr: my function to search the hashes was really efficient
05:17 AM rue_mohr: bah, TM1638 isn't i2c
05:19 AM rue_mohr: er, hmm
05:19 AM rue_mohr: 3am.. hmm
05:25 AM cehteh: antto: i think it should be possible to construct a hash which generates a 8 bit lut for crc8 (after all crc8 has little entropy) .. but i doubt it will be worthwhile, such a hash has to be smaller and/or faster than the lut
05:26 AM cehteh: sometimes it can be, like when your lut is some logarithmic table or some other function and even bettern when its ok that it only returns approximate values
05:37 AM antto: cehteh i've messed with DSP previously, so i've often used math functions for weird curves instead of LUTs
05:38 AM antto: but usually floating point maths
06:13 AM polprog: morning
06:14 AM polprog: https://puu.sh/zTUOs/ab1f229d66.jpg
06:14 AM polprog: probably the dumbest breakout board ever
06:14 AM polprog: i had to solder the pins (single sided) so they dont stick out preventing the soic to be seated
07:19 AM antto: you should've used SMD headers ;P~
07:22 AM Emil: ayyyyyyyy
07:22 AM Emil: My _super_shitty bot control software works
07:22 AM Emil: for telegram
07:23 AM Emil: It's so fragile and so spaghetti that I'm just laguing
07:23 AM Emil: laughing
07:23 AM Emil: Combined it's like 1500 lines of C
07:23 AM Emil: hacked together
07:24 AM Emil: I even started to write my own json parser at 0300 in the morning but decided that a hack was better :D
07:24 AM Emil: ie assume that stuff comes in order and just strstr it :D
07:32 AM polprog: lol
07:32 AM polprog: telegram has a C api?
08:59 AM Ameisen: hmm
08:59 AM antto: hmm^2
09:00 AM Ameisen: pretty sure that I can write set_pins/clear_pins/toggle_pins template functions that will be more efficient than setting/clearing/toggling pins individually in C
09:00 AM Ameisen: since the template can check if any of the pins share the same register, and couple what would be other individual volatile operations
09:00 AM Ameisen: register address*
09:00 AM Ameisen: otherwise*
09:01 AM antto: for avr?
09:02 AM Ameisen: any mcu
09:03 AM Ameisen: they all operate pretty similarly
09:03 AM Ameisen: generally having GPIO pins distributed amongst port registers
09:03 AM Ameisen: most people set up functions like set_pin(PIN_1); set_pin(PIN_2); or such
09:04 AM Ameisen: but if PIN_1 and PIN_2 are on the same PORT register, then that's wasteful
09:04 AM antto: xmega has dedicated registers for set/clear toggle masks
09:04 AM Ameisen: so does ARM
09:04 AM Ameisen: my point still stands
09:04 AM Ameisen: you generally have the masks distributed over multiple registers
09:05 AM Ameisen: but plain functions/volatile ops in C/C++ have no way of concatenating such ops that happen to be on the same register
09:05 AM Ameisen: but a template can
09:09 AM Ameisen: IMO, it's one of the major limitations of 'volatile' - it's overly strict
09:09 AM Ameisen: there really should be arch-specific attributes for 'special' registers like that
09:26 AM Emil: polprog: of course it doesn't
09:26 AM Emil: what do you think I'm writing? :D
09:27 AM polprog: waitwhat.gif
09:27 AM polprog: but it has to communicate to telegrams server some way?
09:28 AM polprog: i dont believe they just made it so anyone can write a simple program and send some data to use it
09:29 AM nohitzzz: https://spin.atomicobject.com/2013/02/08/bit-banding/
09:39 AM Ameisen: bitbanding is a very odd solution to not wanting to include a 'setb/clrb' instruction
09:41 AM Ameisen: Using bitbanding in C an C++ is a bit of a PITA though
09:43 AM cehteh: there was this arch where addresses point to bits too
09:44 AM Ameisen: for 32-bit, all you need is 35-bit pointers
09:44 AM Ameisen: and you can address bits!
09:45 AM Ameisen: Compilers would struggle to take advantage of that
09:47 AM Ameisen: There's just no good reason to have direct bit addressing though. Just provide a dedication instruction for setting/clearing bits in RAM. Sure, it isn't "risc-y", but if you support bitbanding, your CPU already can do it and _is_ doing it, and just doesn't expose an instruction for it.
09:52 AM polprog: compiler is for the weak /s
09:54 AM Ameisen: programmers would, as well
09:54 AM Ameisen: ever write segmented code for x86-16?
09:54 AM Ameisen: ptr+3 addressing would be worse
10:18 AM antto: the xmega also has these virtual ports, i still don't fully understand what they are for, but i think it has something to do with being able to set up a pile of port configurations and then apply all those changes *at once* on a given port
10:37 AM MrFahrenheit: adding bit addresses to pointers seems like a sledgehammer solution
11:10 AM Ameisen: I have to visit the in-laws today
11:10 AM Ameisen: Easter
11:10 AM Ameisen: god her family has so many get-togethers.
11:11 AM Ameisen: and they last so long
11:11 AM Ameisen: and I'm asocial and don't like it
11:12 AM Ameisen: antto - because to use the ports, you don't have to LHS
11:12 AM Ameisen: you are applying settings to them rather than having to read them, alter them, and write them
11:13 AM Ameisen: writing bit 3 to the set port always sets the 3rd pin, and does nothing to the others
11:13 AM Ameisen: thus no read is required.
11:13 AM Ameisen: thus why there are set, clear, and toggle registers
11:13 AM Ameisen: instead of registers that are just the state of the pin
11:13 AM Ameisen: it's more efficient
11:14 AM antto: yes, i was saying that i don't know what the idea for the virtual ports is
11:15 AM Ameisen: not too familiar with xmega, but it appears that the mapped IO ports can use shorter instructions
11:15 AM Ameisen: in/out/sbi/cbi vs lds/sts
11:15 AM Ameisen: since they act like actual registers
11:16 AM Ameisen: seems similar to IO ports on x86
11:16 AM Ameisen: which also have in/out
11:22 AM rue_bed: stm32 is doing that, they have ports for setting/clearing io bits
11:23 AM rue_bed: you can just use normal write insturctions
11:23 AM Ameisen: it's a common design.
11:24 AM rue_bed: it made porting the bitbanged i2c a bit interesting
11:24 AM antto: sure, i bet the xmega didn't invent that, i just didn't understand it when i read about it
11:24 AM rue_bed: I think the same designers work on a lot of the different processors we use
11:24 AM rue_bed: I'v noticed almost identicle subsystem registers
11:25 AM antto: yeah, the NSA
11:25 AM antto: they make sure they got some backdoors in there
11:25 AM antto: undocumented instructions ;P~
11:25 AM rue_bed: *by the way* the port of flurrys i2c for asm on avr to C on stm32 worked
11:26 AM polprog: nice
11:26 AM polprog: good job :P
11:27 AM rue_bed: I'll post it on github, some day other than apr 1
11:27 AM rue_bed: maybe May 4
11:27 AM polprog: heh
11:27 AM rue_bed: :)
11:27 AM rue_bed: I have a pin this year
11:27 AM rue_bed: May the 4th be with you./
11:28 AM polprog: forth ?
11:28 AM rue_bed: meh
11:28 AM polprog: ;)
11:28 AM rue_bed: anything good on /. ?
11:29 AM polprog: i have to wait till wednesday to buy the cd4066 chips
11:29 AM rue_bed: ugh
11:29 AM polprog: so i prototyped the gain stage
11:29 AM polprog: mu first proto worked, a bit
11:29 AM rue_bed: quick you can buy them form aliexpress right now
11:29 AM polprog: my*
11:29 AM rue_bed: they will only take 4 months to show up!
11:29 AM polprog: heh
11:30 AM polprog: on #music-electronics: < Hideki> I don't understand the thinking behind that circuit
11:30 AM polprog: so i changed it a bit per his advice
11:33 AM polprog: now i have moreless this as a gain circuit https://puu.sh/zU3m4/c21bbd6a57.png
11:33 AM polprog: if i had some more parts, read, small audio jacks, i could make a second one like this and then test the summing amplifier
11:34 AM polprog: design a summing amp *
11:34 AM polprog: also need to get double pots
11:41 AM MrFahrenheit: polprog, make a bjt frontend
11:42 AM polprog: hm?
11:42 AM MrFahrenheit: low noise
11:42 AM rue_bed: are you looking for a varialble gian amp?
11:42 AM MrFahrenheit: class a though
11:42 AM polprog: i dont like bjt
11:43 AM rue_bed: I know a part #, I pulled a few from cable tuners
11:43 AM polprog: rue_bed: yeah, a variable gain stage that goes to a mixer. that will be between the mute button and the summing amp
11:44 AM polprog: but... i wanna make it myself, kinda :)
11:44 AM polprog: but sure, if its not a problem, ill happily take a look at the datasheet
11:50 AM polprog: https://puu.sh/zU3VX/7d4dd2c8d5.jpg
11:50 AM rue_shop4: polprog, TDA7439
11:50 AM rue_shop4: check it out
11:50 AM rue_shop4: you can ALSO use a DAC as a digital gian control by putting your signal into the vref pin
11:51 AM unclenorton: is it possible to measure a 3.7k potentiometer without letting more than 10uA through it?
11:52 AM polprog: rue_shop4: looks cool. datasheet says its obsolete though.
11:52 AM rue_shop4: oh, hmm
11:52 AM polprog: im thinking how i could add a simple bass and treble damp/gain EQ
11:52 AM polprog: but thats for way later
11:52 AM rue_shop4: prolly means they want a mint for htem on ali
11:52 AM polprog: possible
11:53 AM rue_shop4: 50c ea
11:53 AM rue_shop4: well 50c-1.00
11:54 AM polprog: maybe a headphone output would be nice
11:54 AM polprog: hmm yeah
11:55 AM polprog: with a separate summing amp! like a monitor output on a real mixing console
11:55 AM rue_shop4: https://www.aliexpress.com/item/Super-micro-digital-power-amplifier-board-2-3W-D-PAM8403-power-amplifier-board-with-high-efficiency/32846713587.html
11:55 AM rue_shop4: I have 10 of those to try out
11:55 AM MrFahrenheit: unclenorton, using a kelvin connection and a good meter, probably - not sure how significant the kelvin connection is though, since you're not measuring a small resistance, but you're still dealing with mV so it might help
11:56 AM MrFahrenheit: you could also use an instrumentation amplifier to make the voltage more measurable
11:56 AM polprog: id use a microampere current source. but those are so small values that a good meter is crucial
11:57 AM polprog: so you measure the actual output of the current source and then the voltage across the pot
11:58 AM rue_shop4: ohm -i 0.00001 -r 3700
11:58 AM rue_shop4: Wattage is: 0.000000
11:58 AM rue_shop4: Current is: 0.000010
11:58 AM rue_shop4: Voltage is: 0.037000
11:58 AM rue_shop4: Resistance is : 3700.000000
11:58 AM rue_shop4: calc 1.25/1024
11:58 AM rue_shop4: 1.25/1024 -->> 0.0012207
11:59 AM rue_shop4: calc .037/.0012207
11:59 AM rue_shop4: .037/.0012207 -->> 30.3105
11:59 AM rue_shop4: so, 30 counts on an adc using a 1.25V ref
12:00 PM rue_shop4: I'd imagine noise would be an issue
12:00 PM unclenorton: MrFahrenheit, i'm talking about measuring it with an avr
12:01 PM rue_shop4: "STOP THINKING SO LOUD, YOUR THROWING OFF THE MEASUREMENT"
12:01 PM MrFahrenheit: use an instrumentation amplifier
12:01 PM rue_shop4: oh, the avr has a built in amp
12:01 PM rue_shop4: what is it, x10?
12:01 PM rue_shop4: I'v never used it
12:01 PM MrFahrenheit: probably shit
12:01 PM rue_shop4: wait, why didn't I use it?
12:01 PM * rue_shop4 has a revelation
12:20 PM polprog: rue_shop4: http://ruemohr.org/~ircjunk/shop/dscn9996.jpg why do you need a punchdown block in your shop ?
12:21 PM polprog: also thats a really nice utility box there :D i wouldnt have thought to put a patch panel inside a kitchen cupboard :D
04:36 PM rue_shop3: :)
04:36 PM rue_shop3: its the shop data closet
04:36 PM rue_shop3: tel/data lines
04:37 PM nohitzzz: nice shop
04:37 PM rue_shop3: :)
04:40 PM polprog: tell you what, sometimes im running out of ports on the wrt54 that serves my bedroom
04:43 PM rue_shop3: http://ruemohr.org/~ircjunk/robots/arm5/p1030984.jpg that data closet @ top left
04:43 PM rue_shop3: http://ruemohr.org/~ircjunk/robots/arm5/p1030724.jpg bit of a shot of the parts wall
04:45 PM rue_shop3: huh, I re-arrange the shop a lot
04:46 PM nohitzzz: was it british columbia?
04:46 PM nohitzzz: nature there looks very much like in finland
04:48 PM nohitzzz: northern finlands
04:56 PM polprog: i peeked through your shop build photos cos im curious about the building side, we may have to build an all-year extension to our summer house
04:57 PM polprog: wanted to look at the insulation/heating part
04:59 PM polprog: rue, tell me your arm is arm based :p
05:42 PM nohitzzz: polprog http://www.deflemask.com/
05:43 PM polprog: looks nice
05:43 PM polprog: shame i dont have luck in composing tracks
05:46 PM MrFahrenheit: oo, big shop
06:05 PM nohitzzz: also goatTracker 2.73 is pretty cool
06:12 PM polprog: next project after the mixer is a robot
06:13 PM polprog: probably gonna stick a raspi with a camera
06:13 PM polprog: on this ex-rc-car base
06:13 PM polprog: it has batteries and mechanical bits
06:13 PM polprog: nice big tyres etc
06:14 PM polprog: beefy dc motor which i would most likely control with a beefier mosfet
06:14 PM polprog: pwming it at several khz so it doesnt cook
06:14 PM polprog: and it would just drive around and transmit picture for starters
06:15 PM nohitzzz: i wrote linux c server side for this and c# windows control client https://www.amazon.com/SunFounder-Raspberry-Robotics-Electronic-Included/dp/B014KK89BW
06:15 PM nohitzzz: with tcp sockets
06:15 PM nohitzzz: it was nice little project for school
06:16 PM polprog: so like xcoffee
06:18 PM polprog: probably gonna add a PPM to my mixer
06:19 PM polprog: possibly on output stage or if i had a rotary switch, on a selectable in/out
06:21 PM nohitzzz: we have been planning for a robot also with a friend, we have a frame for it
06:22 PM nohitzzz: this kind of frame https://ae01.alicdn.com/kf/HTB1Y5_1IFXXXXbqXFXXq6xXFXXX5/T100-tank-car-chassis-caterpillar-track-obstacle-surmounting-with-high-torque-motors-and-hall-sensors-for.jpg_640x640.jpg
06:23 PM nohitzzz: also with raspi, but with windows 10 IOT core, so we could make the robot speak and have speech recognition easily
06:24 PM nohitzzz: so you could ask "Robot, what's the time?" and it would respond
06:25 PM Emil: nohitzzz: still easier on lenux
06:25 PM polprog: import asimov
06:25 PM polprog: also what Emil said
06:25 PM nohitzzz: i doubt it
06:26 PM polprog: stuff like speech recog is so high level you probably gon a use a python/whatever api that is platform agnostic
06:27 PM polprog: windows for me is a PC operating system, any program that doesnt have a window is unusable on it
06:27 PM polprog: imo
06:27 PM Emil: nohitzzz: literally everything that requires any tweaking is easier on lenux, not to mention flexible. Even high level work like python is easier on linux because of the flexibiliy.
06:28 PM polprog: also the whole networking part, once you set it up its gonna last forever
06:28 PM nohitzzz: win10 has apis for speech synthesis/speech recognition
06:32 PM polprog: whatever works
06:36 PM Emil: aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
06:36 PM Emil: fine
06:36 PM Emil: I'll go move my stuff
06:37 PM Emil: even though it's a holiday
06:37 PM Emil: can't trust anyone
06:38 PM rue_mohr: is the synth better than festival?
06:40 PM polprog: heh festival
06:41 PM polprog: installed it one day, ran a default voice
06:41 PM polprog: scottish accent, i nearly fell of my chair laughing :')
06:42 PM nohitzzz: cant say, since i havent used festival
06:42 PM polprog: ive used espeak (steven hawking mode on) or festival
06:43 PM polprog: espeak is more robotic sounding
06:43 PM polprog: festival is just funny
06:43 PM polprog: also https://youtu.be/m8gtdC6-Z14
06:43 PM polprog: ivona synth orders a pizza
06:55 PM polprog: niters
06:55 PM nohitzzz: nite
11:15 PM rue_shop3: festival is garbage
11:15 PM rue_shop3: and it never seems to get better
11:35 PM day__ is now known as day