<awygle>
has anyone seen the video of the JAXA engineers vs. the really old machinists making tops?
<Bike>
i guess poundal and pound force are from different systems
<rqou>
they are?
<rqou>
i never knew/understood that
<pie_>
i dont even find the thought of freedom units amusing, just disgusting :P
<Bike>
i mean why bother knowing it, right?
<pie_>
but something something "human sized units"
<rqou>
also brilliant units like Rankines
<Bike>
pound mass and force are "English Engineering units"
<Bike>
poundals are part of foot-pound-second
<Bike>
and rankines are awesome
<rqou>
at least we aren't weirdos using units like dynes and measuring capacitance in centimeters
<awygle>
here's the thing about english units - if you want to get stuff built in the US, they're important
<rqou>
but who would want to do that when china is so much better? :P
<Bike>
i'm just glad we don't use them in education much anymore
<awygle>
at planetary we tried to do all our PCB stuff in metric but then our fab partners were like "what's a millimeter"
<Bike>
i know there's engineering stuff still inexplicably, desperately hanging on to the shit
<Bike>
but jesus
<azonenberg>
awygle: eeeew
<rqou>
awygle: i dare you to give them a kicad gerber
<Bike>
my high school science teacher convinced us to learn metric by handing out a sheet of imperial conversions
<Bike>
like gallons to hogsheads
<Bike>
very convincing
<awygle>
how big is a 1005?
<rqou>
kicad gerbers are generated in nanometers
<awygle>
(1005 metric is an 0402 english)
<rqou>
apparently they could never get unit conversions to work correctly without causing floating point issues that can cause self-intersections and make polygons non-convex
<azonenberg>
awygle: i use english component size names
<azonenberg>
but metric for all dimensions
<rqou>
^
<azonenberg>
my footprints reflect this
<azonenberg>
EIA_0402_CAP_NOSILK
<azonenberg>
etc
<azonenberg>
makes it very explicit what system i'm using
<awygle>
yeah i mean it makes me sad because 0402 is the "common resistor size" but then we have 0.5mm BSC QFPs etc
<azonenberg>
and 1.27mm SOICs
<awygle>
50mil
<azonenberg>
Exactly
<azonenberg>
(which i ban from my boards unless forced to use them... soic is MASSIVE)
<awygle>
lol
<azonenberg>
I go DFN/QFN as my first choice package for low pin count parts
<azonenberg>
and bga for high pin count
<awygle>
6mm by 5mm is enormous :P
<awygle>
SC-70 is my favorite package i think
<awygle>
(otherwise known as SOT-223 i think?)
<azonenberg>
i go leadless any time i can
<azonenberg>
and tqfp... *shudder*
<awygle>
SC-70 is a great way to solve the "easy rework vs board size" tradeoff
<awygle>
SC-70s are real smol but the pins are still exposed for blue-wiring or probing
<azonenberg>
what's wrong with reworking a dfn?
<awygle>
azonenberg: let's just take our traditional "leaded vs leadless packages" debate as read at this point :p
<awygle>
i will say QFP makes me very nervous when doing two-sided reflow
<qu1j0t3>
/b 5
<awygle>
i always keep all the big leaded packages on the top side
m_w_ has quit [Ping timeout: 260 seconds]
rohitksingh has joined ##openfpga
noobineer has joined ##openfpga
noobineer has quit [Remote host closed the connection]
noobineer has joined ##openfpga
<rqou>
offtopic: why does the word "asymptotically" trigger semantic satiation so quickly?
noobineer has quit [Quit: Leaving]
rohitksingh has quit [Quit: Leaving.]
<cyrozap>
azonenberg: Not sure if you've seen this already, but I thought you might be interested since you were working on some software to talk to your fancy new scope: https://lxi-tools.github.io/
<cyrozap>
azonenberg: The GUI and C-library are separate, so you can use it with your own applications. It also includes device discovery functionality, so you don't have to manually enter the IP for a device.
<awygle>
ooo
<awygle>
oh sweet with mDNS
* awygle
approves of this strongly
<awygle>
and there's just a link to a PDF of the standard, no nonsense! hell yeah
* rqou
stabs class-K and class-KL functions
<rqou>
i hate time-varying systems
unixb0y has quit [Ping timeout: 256 seconds]
<rqou>
awygle how much do you like math? :P
<awygle>
rqou: not that much :P
unixb0y has joined ##openfpga
<awygle>
there's no good chinese food where i live... :(
<rqou>
no asian ghetto? :P
<awygle>
not in redmond. there's some good thai, and 1000 teriyaki places, but no good sichuan chinese food
<awygle>
i am probably gonna go to the new thai place and feel vaguely dissatisfied. on the plus side, i can actually _get_ to the new thai place now so i guess i shouldn't complain :p
<rqou>
finite escape time, how does it work
<rqou>
i hate nonlinear systems
<rqou>
young's inequality is also a nice piece of magic
digshadow has quit [Ping timeout: 276 seconds]
m_w_ has joined ##openfpga
Ultrasauce has quit [Quit: No Ping reply in 180 seconds.]
Ultrasauce has joined ##openfpga
<rqou>
btw azonenberg what do you want to do with my libftdi PR?
<rqou>
it's a huge pile of copypasta
digshadow has joined ##openfpga
pie_ has quit [Read error: Connection reset by peer]
pie__ has joined ##openfpga
rohitksingh_work has joined ##openfpga
<azonenberg>
awygle: but QFNs stay on the bottom much better
<azonenberg>
one more reason to use them
<azonenberg>
cyrozap: there's no lxi afaik
<awygle>
that is true, and one of the _only_ reasons to use them :p
<azonenberg>
cyrozap: i'm speaking VICP over raw TCP to it
<azonenberg>
SCPI over VICP over TCP*
<azonenberg>
then the PSUs are SCPI over raw TCP
<cyrozap>
Ah, I see. xkcd_standard_proliferation.jpg
<rqou>
heh, ẍ + (δ + ϵ cos 2t)x = 0 is a fun equation
<rqou>
everyone try simulating it :P
<azonenberg>
cyrozap: lol exactly
<azonenberg>
well, for my own instruments i'm going to add a new standard :p
<azonenberg>
SCPI was made for decades-old instruments that didn't move a lot of data
<azonenberg>
it's horribly awkward to parse
<azonenberg>
Which is why i'm going with protobufs
<azonenberg>
for control plane
<azonenberg>
and raw packed binary for bulk data
<awygle>
azonenberg: please support dns-sd
<azonenberg>
it's not human readable but if you plan to push 40 Gbps of DSO samples over a single TCP stream
<azonenberg>
i don't want to be maxing out eight cores parsing it
<awygle>
(i do not expect this plea to be answered)
<awygle>
azonenberg: did you think about e.g. cap'nproto?
<awygle>
or another "zero parsing" format
<azonenberg>
awygle: protobuf is a widely used industry standard, and the parsing overhead for control plane data isn't that significant
<azonenberg>
Bulk data is raw binary for sure, thats the only way to do it
<awygle>
hmm i'll give you "widely used" at least lol. nvm
<azonenberg>
if i could do TCP straight to the GPU i would
<azonenberg>
but... no antikernel :p
<azonenberg>
basically, protobuf parsing overhead is significant if you have lots of complex data structures
<azonenberg>
it is not, however, that bad if you are moving a massive "bytes" blob
<azonenberg>
Which is the typical use case for my high bandwidth stuff
<azonenberg>
SCPI by comparison is all ascii all the time
<azonenberg>
(eew)
<azonenberg>
plus the opcodes are variable length ascii and a pain to implement in an embedded state machine
<awygle>
SCPI is bad
<awygle>
although you can often ask for "raw binary" data (at least my scope and SA both support some such mode)
<azonenberg>
oh you think that's bad
<azonenberg>
lol
<azonenberg>
you havent seen the protocol for the logic analyzer on the lecroy i have
<azonenberg>
So, at the lowest level there's a TCP socket
<azonenberg>
The protocol over TCP is VICP, lecroy's custom transport layer
<azonenberg>
(why they don't do SCPI over TCP directly is beyond me)
<azonenberg>
Then there's SCPI on top of that
<azonenberg>
Then there's a SCPI command to tunnel a vbscript command that pokes COM objects
<azonenberg>
Which you use to configure the various properties of the LA
<azonenberg>
Once it triggers there's another SCPI command to dump the LA trace
<azonenberg>
Which is XML
<azonenberg>
meanwhile the waveform data for the oscilloscope channels use a DIFFERENT scpi command, which returns raw waveform data as an array of %f'd ascii floats
<azonenberg>
During all of this the poor cpu on the scope is maxed out doing conversions, and i dont think it even gets close to maxing out the 100baseT link
<azonenberg>
for my own scope i want the ability to saturate 10/40GbE with line-rate traffic
<awygle>
you lost me at COM
<azonenberg>
awygle: lecroy is very windows heavy
<azonenberg>
they make great hardware but eew
<azonenberg>
its not even full windows on this scope, its "windows embedded" which is some windows ce variant
<azonenberg>
(the higher end stuff runs full windows)
<awygle>
tinyfpga: ping?
<awygle>
can i copy a bunch of stuff on a kicad schematic? seems like i can either select and move it or copy one thing at a time
<awygle>
i guess i should ask on #kicad come to think
<azonenberg>
you should be able to copy nad paste
<azonenberg>
after selecting it
<awygle>
select moves right into "move"
<azonenberg>
you should be able to right click and say copy iirc
<awygle>
argh okay
<azonenberg>
once it's selected
<awygle>
train me to just press "C" and then immediately make it CTRL+C
<awygle>
cool cool cool
<azonenberg>
i bet there's a keyboard shortcut
<azonenberg>
but i dont know what it is
<awygle>
it's ctrl c
<azonenberg>
Logical enough
<awygle>
but "copy" when not selected is just "c"
<azonenberg>
lol
<azonenberg>
thats a bit strange
<rqou>
awygle what are you working on?
<awygle>
rqou: hugefpga c3
<rqou>
lol
<rqou>
are you actually using that name?
<awygle>
no
<awygle>
lol
<azonenberg>
awygle: what's it do?
<awygle>
azonenberg: it's the tinyfpga b2 except using an LM part
<awygle>
90% of its purpose in life is "a cheap dev board to add LM support to icestorm"
<azonenberg>
Makes sense
<azonenberg>
that's basically the same purpose as that coolrunner board rqou has
<awygle>
it accomplishes several of my goals, including "finally actually do a board in kicad"
<awygle>
also "do a board at all" and "have a side project to work on that doesn't involve software at all" :p
<azonenberg>
you've never done a pcb?
<awygle>
i've done tons
<awygle>
just not recently
<azonenberg>
oh that makes more sense
<awygle>
getting the itch lol
<awygle>
the last board i did was in June of last year i think
<azonenberg>
My first pcb project after the move and lab is operational
<azonenberg>
will be a respin of reworkctf
<azonenberg>
to fix the various bugs in v0.1
<azonenberg>
then hopefully make a wider launch
<azonenberg>
get scanlime working on it
<awygle>
this board is pretty simple, i'll probably finish it by friday
<awygle>
at some point i need to build my mixer boards and test them with my new sdr
<awygle>
(that's the board i fabbed in June, i've never actually assembled it)
<azonenberg>
i have greenpak characterization boards from the summer
<azonenberg>
never assembled
<awygle>
i even bought RoHS solder :p
<azonenberg>
:)
<azonenberg>
sac305 is the way to go
<azonenberg>
higher surface tension, better for 2-side reflow
<azonenberg>
no compatibility issues with modern bgas
Bike has quit [Quit: Lost terminal]
<rqou>
hurr durr hurr durr i missed a trivial proof by contradiction
<rqou>
and spent at least 30 minutes on it
<awygle>
Lol
<awygle>
"cutepga"
<rqou>
fpg-nya :P
<awygle>
fpg-nyoro~n...
m_w_ has quit [Ping timeout: 240 seconds]
pie__ has quit [Ping timeout: 240 seconds]
<rqou>
goddammit
<rqou>
this other part of this question is a troll too
<cr1901_modern>
Do you have any smoked cheese awygle?
<awygle>
cr1901_modern: you already had some today
<cr1901_modern>
nyoro~n
<azonenberg>
lol
<azonenberg>
what is this, ##openfpga-and-cats?
<cr1901_modern>
Idk about the others. _I'm_ referencing a spin-off to an overrated anime
* azonenberg
turns around for 30 seconds and this channel is full of anime catgirls
<awygle>
I like anime but mostly just fluffy stuff
<cr1901_modern>
azonenberg: So ally is the extreme weeb?
* azonenberg
looks over at shelf full of pokemon plushies next to her desk
<azonenberg>
yeah, i think so
<cr1901_modern>
I like old 80's stuff before the market got saturated and artists were paid so horribly that big-names like DB Super can't get proper animated sequences.
<azonenberg>
cr1901_modern: that plus artist leads to interesting combinations, she painted me a life sized shiny mudkip cardboard cutout shortly after we started dating
<cr1901_modern>
I have no idea what that means awygle
<awygle>
cr1901_modern: "Goodnight"
<rqou>
anybody want to try to figure out why (-y-3/2x^2-1/2x^3)(x-(y-(x-3/2x^2-1/2x^3))(1-3x-3/2x^2))+(-1/2x^3-3x^2-x-y+1)(y-(x-3/2x^2-1/2x^3)) is not negative definite? :P
<cr1901_modern>
Ahhh, I see
<awygle>
("oyasumi nasai")
<cr1901_modern>
I suppose I'm not weeb bastard enough
<rqou>
nobody wants to try and help me debug my lyapunov function? :P
<cr1901_modern>
No, the last time I tried understanding lyapunov exponent I failed miserably
<cr1901_modern>
But that was 2015, maybe I'm smarter now
<rqou>
um, afaict that's something completely different
<rqou>
ah screw it, SIGALRM occurred :P
<rqou>
it's 11:50 pacific time
* azonenberg
notes he has work in the morning
<azonenberg>
but this code isn't working yet...
<cr1901_modern>
rqou: I downloaded OBS, btw
<cr1901_modern>
And set it up*. Looks like I'm set to stream me drawing polygons
<rqou>
btw why were you looking up the lyapunov exponent?
<cr1901_modern>
I needed it to understand a paper back in grad school
<rqou>
O_o a paper about what?
<cr1901_modern>
spoiler: I never did understand said paper
<rqou>
i didn't know you did dynamical systems/controls stuff
<cr1901_modern>
I was trying to characterize a signal for machine learning in 3 different ways- time domain, freq domain, and using dynamical systems to "sum up" the signal as a number (lyapunov exponent)
<cr1901_modern>
actually 4 different ways*- koch dimension. I never did grok the last two
<rqou>
huh, i've never encountered any of these
<cr1901_modern>
Like I said, please keep in mind I never did actually understand the paper, or the concepts really well. Bike (who isn't here atm) pointed me to a good textbook on dynamic systems a long time ago tho.
<cr1901_modern>
I should ask him again if he remembers, b/c I sure don't
Zorix has quit [*.net *.split]
stefanct has quit [*.net *.split]
Marex has quit [*.net *.split]
ondrej2 has quit [*.net *.split]
futarisIRCcloud has quit [*.net *.split]
_florent_ has quit [*.net *.split]
Patater has quit [*.net *.split]
stefanct__ has joined ##openfpga
futarisIRCcloud has joined ##openfpga
ondrej2 has joined ##openfpga
_florent_ has joined ##openfpga
stefanct__ is now known as stefanct
stefanct has joined ##openfpga
stefanct has quit [Changing host]
Zorix has joined ##openfpga
Patater has joined ##openfpga
<rqou>
hmm
<rqou>
azonenberg: on the xc2c32a, it might actually be possible to verify for certain whether the zia has unroutable paths
<azonenberg>
oh
<azonenberg>
?
<rqou>
just by brute force
<azonenberg>
seems like too many options
<rqou>
there are only 32 possible things to select from, right?
<azonenberg>
64
<rqou>
oh nvm
<azonenberg>
ios plus macrocells
<rqou>
right
<rqou>
yeah, that doesn't work
<azonenberg>
that being said, we have the zia mapping, you might be able to do something with a sat solver
<rqou>
can you?
<azonenberg>
or smt
<azonenberg>
Dont know
Marex has joined ##openfpga
FabM has joined ##openfpga
FabM is now known as FabM_Cave
nrossi has quit [Ping timeout: 276 seconds]
nrossi has joined ##openfpga
openfpga-github has joined ##openfpga
openfpga-github has left ##openfpga [##openfpga]
<openfpga-github>
openfpga/master 3a71082 Robert Ou: xc2par: Add some very basic PAR reftests
<openfpga-github>
openfpga/master 210d52c Robert Ou: xc2par: Add binaries to run various PAR stages...
<openfpga-github>
[openfpga] rqou pushed 2 new commits to master: https://git.io/vxW6O
<rqou>
azonenberg: guess what
<rqou>
the code i wrote months ago is actually complete enough for a blinky on coolrunner2
<rqou>
except due to a number of issues it doesn't blink at a human-visible rate
openfpga-github has joined ##openfpga
openfpga-github has left ##openfpga [##openfpga]
<openfpga-github>
openfpga/master 7fd4b60 Robert Ou: xc2par: Separate out unconstrained p-terms to greedily assign them
<openfpga-github>
[openfpga] rqou pushed 1 new commit to master: https://git.io/vxWSl
m_w_ has joined ##openfpga
pie__ has joined ##openfpga
m_w_ has quit [Ping timeout: 268 seconds]
pie__ has quit [Ping timeout: 256 seconds]
eduardo__ has joined ##openfpga
eduardo_ has quit [Ping timeout: 240 seconds]
pie__ has joined ##openfpga
Bike has joined ##openfpga
<Bike>
cr1901_modern: not sure. i probably got the most out of "Dynamics: The Geometry of Behavior", which is a picture book and written by a crazy person
<Bike>
otherwise there's like, pick one by arnol'd
<Bike>
or a neuroscience book i liked
rohitksingh_work has quit [Read error: Connection reset by peer]
<Bike>
as far as i know most of the theory amounts to "if you can find a place where the system is sort of linear, do that. otherwise go fuck yourself", which works with lyapunov exponents
<cr1901_modern>
The picture book was the one I was thinking of if memory serves
<cr1901_modern>
And hey, "if you can find a place where the system is sort of linear, do that. otherwise go fuck yourself" you just described electronics :D
<Bike>
yeah, about halfway through my intro to electronics class i realized it was just hartman-grobman for people who don't know dynamical systems
<Bike>
which is good because i spent a long time going "the MOSFET is linear, you say. hm [stares at iv curve] hmmmmm"
<cr1901_modern>
Try solving for the exact current through a series diode and resistor attached to 1V
<Bike>
uhhhh do you need a lambert w there
<cr1901_modern>
I can't say I know what that is. The answer I was expecting was "no closed form solution" :)
<Bike>
i was mostly pretty confused because earlier i'd read a book on electronics that used subthreshold bjts to do arithmetic, and i thought that was normal
<Bike>
z = W(z*exp(z)), so it comes up when you try to invert things with exponents sometimes
<sorear>
I…don't think there's a closed form for the I-V characteristic of real diodes? Series resister not necessary
<Bike>
isn't it just the shockley equation
<cr1901_modern>
Shockley equation
<Bike>
oh REAL diodes
<Bike>
i mean yeah it's not going to have breakdown and shit
<Bike>
or you could do some piecewise defined thing
<Bike>
i remember looking into how tools treated transistor i-vs once, it wasn't pretty
<cr1901_modern>
SPIRCE uses Gummel-Poon model
<cr1901_modern>
SPICE* even
<Bike>
i like how wikipedia just diagrams the circuit, lists 41 parameters, and says good luck
<cr1901_modern>
sorear: In any case, I meant ideal devices in my toy example. You can solve for the current thru a diode by itself using Shockley equation, but once you add a resistor there's no closed form solution.
<cr1901_modern>
The 1970s must've been wild times to come up with that monstrosity
<sorear>
just make something up
<sorear>
you're already using a fake formula for the diode, make a fake formula for the diode+resistor
<Bike>
i mean it would probably be piecewise
<Bike>
or a non-elementary function
<Bike>
at that point you should just like, specify it as a D.E. and go from there. only using this shit for computation anyway
<sorear>
all of this is characterization problems anyway
<Bike>
cr1901_modern: btw note that the picture book costs like five hundred dollars for some reason
<Bike>
if you find it online anywhere plz share
<cr1901_modern>
Will do
<cr1901_modern>
Could've sworn I saw it online tho
<cr1901_modern>
well, back then anyway*
<Bike>
maybe
<Bike>
i shoulda scanned it
<Bike>
though i think the library didn't have all the volumes
<Bike>
oh hey my work library has it
<cr1901_modern>
Maybe we'll have it online soon enough (tm) :D?
<Bike>
as well as other books by the author, like "Chaos, gaia, eros: a chaos pioneer uncovers the three great streams of history"
<cr1901_modern>
Sounds like bad scifi
<Bike>
dsoesn't seem to have his book on DMT. sad
pie__ has quit [Ping timeout: 248 seconds]
<azonenberg>
rqou: lol
azonenberg_work has quit [Ping timeout: 252 seconds]
Bike_ has joined ##openfpga
Bike has quit [Disconnected by services]
Bike_ is now known as Bike
<Bike>
mhm, cas can give the current through a diode in series with a resistor using the lambert w function
<Bike>
form: closed* (*depending on what you mean by)
<Bike>
making wolfram alpha recognize algebra is such a pain in the ass
<cr1901_modern>
Interesting I'll have to try that/learn about lambert w
<cr1901_modern>
By closed, I meant informally "using add sub mul div log exp manip"
<cr1901_modern>
to get "i" in terms of "v"
<Bike>
yeah probably "elementary function" is what you have in mind
<Bike>
no, wait, without arbitrary roots
<cr1901_modern>
I think roots == "exp", just raising to a fractional power
<cr1901_modern>
But yea, I think you're right "elementary function" is what I mean :)
<Bike>
https://i.imgur.com/F59PGcC.png this is what it looks like. S is saturation current, r is the resistor, N is nV_T, for aforementioned pain in the ass reasons
<Bike>
it works for any branch of lambert i guess, weird
<Bike>
probably not really weird
<Bike>
and x is the current. i really need to get an actual cas at some point
kristianpaul has quit [Quit: Lost terminal]
pie_ has joined ##openfpga
<azonenberg>
rqou: btw what are the missing features?
<azonenberg>
i fail to understand how you can have a t flipflop but not chain them until you get single digit Hz
m_w_ has joined ##openfpga
_whitelogger has joined ##openfpga
azonenberg_work has joined ##openfpga
whitequark has joined ##openfpga
azonenberg_work has quit [Ping timeout: 260 seconds]
pie_ has quit [Ping timeout: 276 seconds]
m_w_ has quit [Quit: Leaving]
pie_ has joined ##openfpga
m_w has joined ##openfpga
soylentyellow has quit [Ping timeout: 240 seconds]
azonenberg_work has joined ##openfpga
<pie_>
that sounds like something a bad person would do
<awygle>
not sure there's really an alternative to that
<awygle>
"brick" here isn't "brick-brick", it just means you can't reprogram the FPGA from within the FPGA. which is the normal state for most FPGAs.
<awygle>
all these cool meetups in SF are almost enough to make me wish i was there... but then i remember soaring housing prices and zucchini NIMBYs and come to my senses
<rqou>
zucchini?
<rqou>
also you can always sleep on one of my couches :P
<awygle>
(just what has it on my mind today, not where I first saw it)
<rqou>
lol wtf
<awygle>
i 'preciate that :p planes are sadly expensive tho
<awygle>
speaking of when are you going to come up here for an openfpga hackathon?
<rqou>
drive? :P
<rqou>
awygle: i need to get pointfree or somebody to want to come with me because I don't want to be driving for 14 hours straight
m_w has quit [Quit: leaving]
<awygle>
well you're welcome on my couch anytime :p the cats would enjoy a new jungle gym
<rqou>
yeah, i stayed with azonenberg last time
Bike has quit [Ping timeout: 260 seconds]
<awygle>
yeah tbh he's got like... rooms
<rqou>
was it you who was pointing out that azonenberg is adulting very well? with a job, wife, _and_ house?
<awygle>
lol i did say something like that awhile back
<whitequark>
azonenberg is annoyingly well-adjusted >_>
<rqou>
now he just needs a lawn, and a dog, and 1.84 kids, and a pension, and America will be GREAT again :P :P
<sorear>
ugh
<awygle>
yeah people like azonenberg make it hard to maintain the "i don't have those things because i _choose_ to focus on my _career_" ego-defense
<awygle>
see also: really athletic programmers :p
<whitequark>
eh, not everyone can be like djb or azonenberg
* sorear
would find walking in suburbia more tolerable if there weren't dogs in half of the yards
<sorear>
…does this channel know djb
<cr1901_modern>
I wish I didn't
<whitequark>
who doesn't know djb?
<sorear>
possible there's a different djb under discussion? or is the cryptographer and close associate of jacob appelbaum relevant to azonenberg's life somehow
<whitequark>
i don't know how appelbaum is related, but i was talking about the ridiculously productive cryptographer, yes
<rqou>
appelbaum apparently turned into a grad student under djb
<balrog>
Looks like a speed optimized implementation
<mithro>
pointfree: not talking about anything related to openfpga / symbiflow
<whitequark>
yes but
<whitequark>
194 register variables?!
<cr1901_modern>
isn't register storage class useless anyway?
<sorear>
is there any current compiler where `register` is not a no-op?
<awygle>
yeah i was gonna say, whitequark you know compilers, isn't register basically ignored?
<whitequark>
yes it is
<awygle>
not that that makes it better, really
<whitequark>
precisely
<awygle>
also comments are apparently for losers
m_w has joined ##openfpga
<sorear>
look at least it's not a perl script which generates assembly
<awygle>
hey whitequark, apropos of nothing really, but what vid:pid are you planning to use for Glasgow-JTAG?
<whitequark>
awygle: I'll request something from openmoko
<rqou>
so one thing that i've occasionally thought about writing was some kind of "libpedagogicalcrypto" to explain to "normal programmers" how to do "weird fancy math"
<awygle>
mk, cool
<rqou>
but i don't really know who would be the audience for such a thing
<whitequark>
awygle: what about it
<rqou>
whitequark: not squatting 0xf055?
<sorear>
i've had to explain to people that not having a carry flag does NOT make constant time bignums impossible
<awygle>
whitequark: just wondering. thinking about a future usb project that will need to solve that problem.
<whitequark>
rqou: nope
<awygle>
was wondering if there was a Lattice VID or something
<awygle>
which would be kind of weird
<rqou>
any particular reason why not? other than "i cannot infrastructure for shit"?
<rqou>
awygle: squat 0xf055 and be friends with me :P
<whitequark>
rqou: because it's squatting?
<rqou>
so?
<awygle>
rqou: one out of two ain't bad
<whitequark>
and openmoko has a perfectly legitimate vid
<rqou>
if you don't use USB trademarks they can't do anything about it
<awygle>
didn't openmoko get some kind of directive not to sublet PIDs?
<rqou>
that was one of the other ones
<whitequark>
i don't see any benefits in squatting and i see drawbacks
<whitequark>
why should i use it
<awygle>
oh right that was the "crowdfund a vid" project
<rqou>
they told the USB-IF that rescinding their VID is illegal in the EU
<rqou>
so USB-IF revoked their VID anyways
<rqou>
but they can't reassign it
<rqou>
so effectively nothing happened :P
<pointfree>
mithro: but it does involve fpga hardware?
<awygle>
i can see a long-term benefit to the squatting approach in a kind of signalling way. "hey you bastards stop artificially locking shit out for no reason". but on balance i'm with whitequark on this one.
<mithro>
Yes
<sorear>
why is the USB-IF so weird about this
<awygle>
sorear: $$$
<rqou>
you can also squat 0xf055 for pci though
<rqou>
you can blame the PCI-SIG for that one
<sorear>
is PCI also funded by selling VIDs?
<mithro>
100 slides for a 30 minute talk sounds fine right? :-P
<rqou>
hmm whitequark i don't see any real drawbacks in squatting a VID?
<rqou>
i mean, there's tons of chinese crap using bullshit vid/pids and they still work
<awygle>
well, they're in china. you're in the us.
<whitequark>
rqou: "there's tons of C code invoking UB and it still works"
<whitequark>
it's sloppy.
<rqou>
awygle: as far as i can tell, the only weapons the usb-if have are trademarks and patents
<rqou>
and if you aren't designing silicon or anything like that you probably don't hit patent issues
<whitequark>
this is like releasing software without a license
<rqou>
so as long as you're careful with trademarks you should be fine
<whitequark>
technically, everyone can still use it
<mithro>
That is a whole ~18 seconds per slide.....
<sorear>
if I ever get stuck designing a bus protocol, how does EUI-48 for type identifiers sound. let someone else with a track record of affordability handle the problem
<awygle>
the risk is that they can crush you with a lawsuit. or at least, they could crush *me*.
<whitequark>
also, i don't want to "be careful" with trademarks
<pointfree>
mithro: If it's flip-book stop motion animation hahas
<rqou>
whitequark: but if you didn't sign an agreement with the usb-if you can't use their trademarks anyways
<rqou>
so there's nothing new there
<whitequark>
hrm
<awygle>
rqou: my attitude is, they don't have to _win_ a lawsuit. anybody getting sued by the usb-if is in a bad way. the more you antagonize them the more likely that is to happen.
<rqou>
so as far as i can tell, using an openmoko vid is exactly the same as squatting a vid legally
Bike has joined ##openfpga
<rqou>
i mean, they can always decide to sue you anyways
<whitequark>
how so? openmoko is permitted to suballocate pids
<whitequark>
since it got the vid before the rules changed
<rqou>
i thought this was the "usb-if says rules changed, EU orgs said that's illegal" situation?
<whitequark>
i believe that the agreement between openmoko and usb-if didn't change
<whitequark>
only new ones did
digshadow has joined ##openfpga
<mithro>
Just because you are right doesn't mean that people can't sue you
<rqou>
right, but getting sued is always a risk anyways
<pointfree>
semi-OT: I know that the FCC exempts "professional test equipment" such as the HackRF from compliance testing. But will EU require compliance testing (will I need a Chinese Export CE sticker)?
<rqou>
the usb-if tried to retroactively change the agreement
<rqou>
pointfree: wait, i didn't think there was any such FCC exception?
<whitequark>
there is!
<rqou>
i thought you had to claim yourself as a subassembly?
<rqou>
oh huh there is indeed
<rqou>
"A digital device used exclusively as industrial, commercial, or medical test equipment."
<rqou>
does a devboard fit this definition?
<rqou>
i _think_ a devboard should fit the "subassembly" definition instead?
<rqou>
of course, this is all here in freedomland, afaik the EU cares more about actually having rules that don't have loopholes
<rqou>
also, if i'm reading it right, a subassembly cannot include an enclosure or power supply
<pointfree>
Also, I don't think you need to prove you are a "licensed professional" in any way to upon purchasing a HackRF. The HackRF webpage just says "you are responsible for using this equipment legally."
<rqou>
so (IANAL) it seems you can sell a raspi or a raspi enclosure and psu but you technically can't sell a raspi, enclosure, and psu bundle
<pointfree>
It would be rather difficult to test a software defined radio "in every possible configuration"
<rqou>
yeah, afaik it's definitely on the purchaser to operate the device correctly
<rqou>
see e.g. the flood of baofeng radios
<rqou>
or ham radios that only have part15 certification
<awygle>
yeah SDRs are all under the test equipment thing
<awygle>
don't you have to meet the "unintentional radiator" standard though?
<awygle>
(subassemblies do not)
<rqou>
does what have to meet the "unintentional radiator" standard?
<awygle>
SDRs/test equipment
<rqou>
afaik all the "real" test equipment tests for part 15 unintentional radiator class A
<rqou>
(for business/industrial rather than residential class B)
<awygle>
right but idk if the HackRF is tested to that standard
<rqou>
i believe they don't _have_ to if they're _exclusively_ test equipment
<awygle>
my extensive testing (N=1) says that 15 A is not a hard standard to meet
<pointfree>
My HackRF has a CE marking on it but no FCC marking that I could see.
<rqou>
CE is totally different because europeans don't seem to like loopholes
<rqou>
anyways, i think the idea is that ham stuff or baofeng radios aren't _exclusively_ test equipment so they test for part 15 unintentional radiator class A for the receive functionality
<rqou>
making sure TX follows the rules is on the user
mumptai has quit [Quit: Verlassend]
<awygle>
some quick reading suggests they are using "dev board in a lab" exception to avoid unintentional certification
<rqou>
yeah, i think they can do that
<rqou>
falling under the "exclusively test equipment" exemption
<pointfree>
I don't think ambient backscatter radio requires any testing whatsoever. I don't even know if the FCC mentions them. They are not emitters. How could the FCC tell me when and when not to listen to NPR?
<rqou>
er, you _might_ still fall under the unintentional radiator rules
<whitequark>
that's actually the reason for the cellphone ban in aviation
<whitequark>
shitty heterodyne receivers
<whitequark>
of course that was like 25 years ago
<rqou>
or you can just have a rusty copper pipe and get the same problem :P
<rqou>
CuO is a semiconductor :P
<rqou>
er, *Cu2O
<awygle>
the FCC certifies everything electronic, to some level. everything radiates. that's why it's important to specify what certification you're talking about. "intentional radiator" == "radio transmitter", which "listen to NPR" would not fall under.
<awygle>
"unintentional radiator" == "any electronic device" (to first order), which "listen to NPR" _would_ fall under/
<rqou>
actually, "listening to NPR" _might_ be exempt depending on what you use to listen
<rqou>
i think a crystal detector would be exempt
<rqou>
"Digital devices in which both the highest frequency generated and the highest frequency used are less than 1.705 MHz and which do not operate from the AC power lines or contain provisions for operation while connected to the AC power lines."
<awygle>
isn't a crystal detector analog... :p
<rqou>
oh wait
<rqou>
wtf
<rqou>
uh...
<rqou>
awygle: yeah, this is interesting
<rqou>
i actually have no idea what category a crystal detector would fall into, but afaict a battery-powered AM-only heterodyne receiver should actually fall into the exeption
<pointfree>
an ambient backscatter device has no power source of its own not unlike a piece of aluminum foil ...but at what point would adding a battery or power over usb make it a regulated emitter? Maybe the powered portion needs to be isolated from the transciever?
<rqou>
i think power over usb is bad because it allows you to power it from an ac power source
<rqou>
a battery i think is fine
<rqou>
hmm, if i'm reading it correctly, a purely analog device with no digital techniques at all (e.g. a crystal radio) might just not fall under the equipment marketing rules at all
<sorear>
why does analog vs digital matter here?
<rqou>
i don't know, ask the FCC
<sorear>
analog intentional radiators exist
<rqou>
i think any clock driver circuit would make it fall into the category of "digital device"
<sorear>
heterodyne receivers predate digital anything
azonenberg_work has quit [Ping timeout: 240 seconds]