enriq has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
wpwrak has joined ##openfpga
pie_ has quit [Read error: Connection reset by peer]
pie_ has joined ##openfpga
sgstair has quit [Quit: .•«UPP»•.]
sgstair has joined ##openfpga
abetusk has joined ##openfpga
abetusk is now known as Guest63880
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined ##openfpga
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined ##openfpga
digshadow has quit [Ping timeout: 258 seconds]
<rqou>
hey whitequark, azonenberg: i have a fun side/distraction project: a not-shitty hotplate stirrer (for chemistry)
<cr1901_modern>
Random thought: I would love an automatic stirrer for doing more inane things like autocooking spaghetti
<azonenberg>
magnetic stir spaghetti? lol
<cr1901_modern>
azonenberg: Of course it would need to be sanitary :P. And just an idle thought
<awygle>
actually i'm with cr1901_modern, that would be an improvement to my life
<cr1901_modern>
spaghetti is a cooked food that is A. delicious, and B. easy enough that it could prob be automated
<azonenberg>
lol yeah
<awygle>
except for macaroni and cheese... mmmmm
<rqou>
you can start with an IoT sous vide cooker
<rqou>
(yes, there's one in this apartment)
<rqou>
(no, the IoT part isn't enabled)
<awygle>
IoT crock pot must be a thing
<cr1901_modern>
Hmmm, something to start with...
<cr1901_modern>
Ya know, it's quite a pity... IoT to start cooking/come home to a cooked meal would actually be nice
<cr1901_modern>
but then the world ruins it
<azonenberg>
IoT has become "let's go data-mine our customers and sell it to advertisers"
<rqou>
btw "start with" = "you can just order this off of amazon as a commercial product already"
<azonenberg>
there was so much potential there
<awygle>
seems like that's the only business model in tech :(
<azonenberg>
and advertising wrecked it
<azonenberg>
As they did most of the rest of tech :p
<azonenberg>
i'm waiting for the advertising bubble to burst
<rqou>
when my housemate bought it, i immediately replied "buy one? we have _how many_ berkeley engineers here? we should build one!"
<awygle>
that's.... i'm not sure that's a thing, azonenberg
<rqou>
and the reply was "shut up, <other housemate> already said that exact same thing"
<azonenberg>
awygle: i want to have enough faith in humanity
<azonenberg>
to not think that people are so trivially swayed by advertising
<whitequark>
it isn't necessary to have faith in humanity
<whitequark>
for the ad bubble to burst
<rqou>
"how many berkeley EECS grads does it take to not succeed in building a sous vide cooker?" :P
<azonenberg>
and that as adblockers get easier to use and reach near 100% saturation
<awygle>
i cannot believe that advertising works but i also cannot believe that such a huge industry could be based on something that didn't work. so.
<azonenberg>
eventually ad-supported business models will just stop working
<azonenberg>
awygle: yeah
<cr1901_modern>
To make it perfectly clear: I don't use an adblocker b/c I want to
<whitequark>
awygle: yet it is
<cr1901_modern>
I use one b/c I have to
<whitequark>
it's subsidized by venture capital
<whitequark>
for now
<azonenberg>
i mean, my reaction to seeing an ad is typically "eew, let me not buy from that company"
<whitequark>
at some point, VCs will see that they aren't seeing the gains
<whitequark>
and... pop
<azonenberg>
and then i either hold off on my purchase or buy from a competitor
<awygle>
advertising isn't like, a new thing though
<whitequark>
it isn't, but advertising as we know it right now is
<awygle>
Big Data (TM) is, arguably, but i'm not convinced there is an advertising bubble
<whitequark>
i.e. with hundreds of middlemen
<azonenberg>
i.e. when i see an ad i have an instant, strongly-negative reaction to it
<azonenberg>
i dont go so far as to blacklist the advertisrer from ever getting my business as i'd never be able to buy anything :p
<cr1901_modern>
"Advertising is the arresting of human intelligence long enough to quander money from it"
<awygle>
hey i'm right there with both of you but i'm not sure my aunt susan is
<azonenberg>
but i at least try to avoid giving someone business because of ads
<azonenberg>
cr1901_modern: And gambling...
<azonenberg>
A tax on people who don't understand statistics
<cr1901_modern>
True, but that's not the quote I bastardized
<awygle>
i hope you're both right tho
<rqou>
back on the topic of chemistry stuff, i'm rewatching old nurdrage videos (and procrastinating) and saw the one about platinum electrodes
<rqou>
apparently the fake platinum electrodes that nurdrage got cost about $20 on ebay, but i can just buy a 1g platinum ingot for only about $30
<rqou>
is there any reason to not just do that? :P
digshadow has joined ##openfpga
* DocScrutinizer05
waits for advent of negative ads, tailored to the negative reaction of target customers like azonenberg ;-)
<DocScrutinizer05>
A dioes maximum annoying ads in name of competitor B
<rqou>
i figured that would probably exist already? :P
<DocScrutinizer05>
in politics evidently exists already, yes
<rqou>
"but her emails!"
<DocScrutinizer05>
you think that stopped some people from voting for Trump?
<rqou>
er, "but her emails" was a pro-trump slogan
<DocScrutinizer05>
exactly
<azonenberg>
DocScrutinizer05: people voted for trump or hillary, IMO, to prevent the one they saw as being ultra-horrible from winning and instead having the less-horrible
<DocScrutinizer05>
so for a negative ad it would have to act against him
<azonenberg>
Most people i've talked to, no matter where they fall on the political spectrum
<azonenberg>
agree with my assessment: trump would have lost to any democrat other than hillary
<azonenberg>
and hillary would have lost to any republican other than trump
<azonenberg>
The only chance either had was to run against the other
Guest63880 is now known as abetusk
<cr1901_modern>
"and hillary would have lost to any republican other than trump" And she lost to trump anyway b/c $redacted
<azonenberg>
Sure, but if she had run against anyone else i think she would have lost by a much greater margin
<azonenberg>
Both had among the lowest approval ratings in history at the time of the election iirc
<cr1901_modern>
I want out of this timeline
<DocScrutinizer05>
true negative double-trickery probably was in austrian elections recently, where a campaign designer hired by the left set up a pathetic demeaning wensite about the right candidate, which in turn fell on the feet of the left candidate, but now might turn out the left campaign manager been a mole paid by the right candidate about whom he made that terrible website
<awygle>
i don't think hillary-as-hillary mattered much. i don't think any other democrat-democrat would have had a significantly different campaign. personally.
<whitequark>
rqou: platinum wire maybe
<rqou>
ooh i forgot you can also buy that :P
Bike has quit [Quit: Lost terminal]
jhol has quit [Quit: Coyote finally caught me]
nrossi has joined ##openfpga
pie_ has quit [Remote host closed the connection]
pie_ has joined ##openfpga
cr1901_modern has quit [Ping timeout: 248 seconds]
<azonenberg>
i know i have to add mask testing still
<azonenberg>
as well as either improving the UI measurement or adding a manual override, since it often is off by a bit
theMagnumOrange has quit [Ping timeout: 240 seconds]
<rqou>
oh btw azonenberg fun story about mask testing
<azonenberg>
oh?
<rqou>
a while back a friend asked me "hey do you know how to <basically do a mask test> using <keysight scope>"
<rqou>
and my answer was "does it have the mask test feature enabled?"
<rqou>
and i realized "of course not, none of us would bother with that because we all know that it would be 'easy' to hack up mask testing in python instead"
<azonenberg>
lol
<rqou>
turns out that all the "talk with scope" protocols aren't anything close to "easy" :P
<azonenberg>
Its not so bad once you get into it
<rqou>
my scope needs some stupid "tunnel over sunrpc" think
<rqou>
*thing
<azonenberg>
...
<azonenberg>
oh
<azonenberg>
mine is SCPI commands over a fairly straightforward TCP-based framing protocol
<rqou>
which might be easier to implement completely in python than to actually use any proper sunrpc daemons
<azonenberg>
not sure why they didn't just telnet SCPI over TCP
<rqou>
they did
<rqou>
there's a sunrpc layer in the middle :P
<azonenberg>
No i mean, on my scope
<azonenberg>
VICP is a lecroy custom protocol between TCP and the SCPI commands
<azonenberg>
its not sunrpc
<azonenberg>
Anyway the hard part for me is that lecroy's documentation for the various features of the scope is... lacking
<azonenberg>
some commands are well documented
<azonenberg>
like, how to download a waveform from the scope
<rqou>
aah i figured out what my scope's protocol is called
<rqou>
VXI-11
<whitequark>
rqou: tunnel over *what*
<azonenberg>
that's a pretty standard one but i havent used it since my scope doesnt support it
<azonenberg>
rqou: anyway, its documented pretty well how to get a "waveform" object
<rqou>
ironically it's probably easier on my scope to download a waveform via http than via the VXI-11 protocol :P
<azonenberg>
now what that object MEANS? thats a whole other question :p
<azonenberg>
after a lot of hunting in docs for older lecroy scopes i found a spec for the wavedesc object
<rqou>
oh yeah, my scope also has this random http thing
<rqou>
that also accepts SCPI commands
<rqou>
and can take screenshots
<azonenberg>
which is a packed binary blob that has a bunch of metadata like sample rate and vertical gain/offset
<azonenberg>
So now i can decode everything i care about
<rqou>
afaik my ancient tek literally just sends 10k samples
<azonenberg>
Then a lot of the fun things like protocol decodes or the DVM / function generator seem to be mostly accessed over DCOM
<azonenberg>
although they have some kind of VB parser over SCPI
<rqou>
wat
<azonenberg>
so i can probably access them through that rather than writing DCOM code
<rqou>
double wat
<azonenberg>
The scope runs windows and is heavily COM based internally
<rqou>
all scope protocols are terrible
<azonenberg>
on a windows client you can probably write code to talk to it easily over .net
<azonenberg>
but from linux the best option seems to be SCPI over VICP
<azonenberg>
and yes, they are
<rqou>
so my old troll of fpga-accelerated DCOM isn't totally useless :P :P
<azonenberg>
When i build myself instruments with LAN interfaces they'll be sane
<azonenberg>
As in, one TCP socket for a simple binary control protocol then another that streams out capture data
<azonenberg>
clean control-data plane separations
theMagnumOrange has joined ##openfpga
m_t has joined ##openfpga
pie___ has joined ##openfpga
pie_ has quit [Ping timeout: 240 seconds]
eduardo__ has joined ##openfpga
eduardo_ has quit [Ping timeout: 240 seconds]
qu1j0t3 has quit [Ping timeout: 260 seconds]
qu1j0t3 has joined ##openfpga
enriq has joined ##openfpga
xdeller_ has quit [Ping timeout: 260 seconds]
pie___ has quit [Ping timeout: 252 seconds]
pie_ has joined ##openfpga
pie_ has quit [Remote host closed the connection]
pie_ has joined ##openfpga
jhol has joined ##openfpga
m_w has joined ##openfpga
digshadow has quit [Ping timeout: 240 seconds]
pie_ has quit [Ping timeout: 248 seconds]
clifford has quit [Quit: Ex-Chat]
digshadow has joined ##openfpga
clifford has joined ##openfpga
pointfree1 has quit [Ping timeout: 255 seconds]
Xark has quit [Ping timeout: 240 seconds]
Xark has joined ##openfpga
Hootch has joined ##openfpga
Hootch has quit [Client Quit]
Hootch has joined ##openfpga
pie_ has joined ##openfpga
nrossi has quit [Quit: Connection closed for inactivity]
Hootch has quit [Quit: Leaving]
fpgacraft1 has quit [Quit: ZNC 1.7.x-git-709-1bb0199 - http://znc.in]
<rqou>
maybe i should just do what some people do and just make it an achievement to be on all the watchlists (except the "SSSS" list, that one sucks) :P
uelen has quit [Ping timeout: 248 seconds]
<rqou>
azonenberg_work: somewhat serious suggestion: webgl port for your waveform thingy?
<azonenberg_work>
Not a chance
<rqou>
why not?
<azonenberg_work>
although an OpenGL native port could happen
<azonenberg_work>
Because javascript
<azonenberg_work>
i actually am seriously considering a native GL port
<rqou>
is it not gl-rendered?
<rqou>
what does it use right now?
<azonenberg_work>
cairo
<rqou>
ah
<rqou>
if you do a gles port, i can make try to make it webified
<azonenberg_work>
i have a lot of UI work to do in general
<cr1901_modern>
"Because javascript" <-- like it or not, azonenberg, javascript and browsers won the UI wars. :/
<azonenberg_work>
cr1901_modern: no, they won the UI wars for trivial things that dont deal with a lot of data or high performance and are used by filthy casuals :p
<azonenberg_work>
nobody is porting Paraview to javascript
<cr1901_modern>
We can't all be sublime text
<cr1901_modern>
where you say f*** it and just draw all widgets to a framebuffer
<azonenberg_work>
I'm actually looking at moving in the opposite direction lol
<azonenberg_work>
The 100M ethernet and crappy CPU on my lecroy arent fast enough to push the thousands of WFM/s that i want
<azonenberg_work>
My homebrewed test equipment will have multiple lanes of 10G or even 40G Ethernet on it
<azonenberg_work>
Using a very low overhead binary protocol
<azonenberg_work>
this just isn't something you do in javascript
<azonenberg_work>
the object metadata overhead alone is probably enough to 10x your RAM
<azonenberg_work>
awygle: To give you an idea of where i'm looking at taking this project long term
<azonenberg_work>
a starshipraider with 4GB of DDR3 as a capture buffer
<azonenberg_work>
with 32 LA channels
<azonenberg_work>
will be able to record 1B points without compression
<azonenberg_work>
As in 1E9 samples
<azonenberg_work>
At 500 MSa/s that's two seconds of nonstop capture
<azonenberg_work>
Alternatively, your raw capture rate is 16 Gbps so if you bonded two lanes of 10G Ethernet you could do *realtime* capture to a PC with single-digit μs of latency
<azonenberg_work>
And if i only end up having a single lane you'd have to either buffer and stream off (lowering your WFM/s a bit) or lower your sample rate/channel count a bit
<azonenberg_work>
Point is, i want a tool that's able to actually handle that kind of data rates eventually
<azonenberg_work>
I'm probably going to have to push raw waveform data to the GPU and then use a vertex shader to do the gain/offset math
<azonenberg_work>
etc
<rqou>
but you can still have a webgl waveform viewer
<azonenberg_work>
this isnt a waveform viewer, this is a whole oscilloscope UI
<azonenberg_work>
the interactivity is key
<azonenberg_work>
writing just a viewer would involve ripping half the functionality out
<azonenberg_work>
and probably rewriting a lot of the rest
<rqou>
hmm, but i still think a "lite" version to show people a capture would be useful
<azonenberg_work>
or they could, you know, download and install the analysis app and download the waveform file
<azonenberg_work>
or, alternatively, connect directly to the actual scope over VPN/LAN