digshadow has quit [Ping timeout: 264 seconds]
ZipCPU has quit [Ping timeout: 264 seconds]
ZipCPU has joined ##openfpga
<whitequark> awygle: care to poke at any other footprint while i debug this?
<qu1j0t3> whitequark | "nice" would be addressing root causes <-------!!!
* qu1j0t3 was telling someone about that affair the other day
<whitequark> i mean, for all i know they just murdered some kinda goon who was trying to survive, and a next one will immediately take their place
* qu1j0t3 nods
<whitequark> there might be a case for death penalty but this is not it
<qu1j0t3> it seems that it was also a textbook (also horrific) example of Goodhart's Law?
<whitequark> well, not entirely
<qu1j0t3> which at the time i wasn't aware of
<whitequark> you *could* test for protein with a mass spectrometer
<qu1j0t3> (the law i mean)
<qu1j0t3> (nice to have a name for it)
<whitequark> this would allow you to not only detect adulteration with melamine but also with foreign protein
<whitequark> hopefully making it too expensive to do any of it
<awygle> whitequark: gonna do the level shifter symbol in a second here, drugging cat
<whitequark> \o/
<qu1j0t3> whitequark: I suspect a proper protein test would move the shenanigans elsewhere. See also: https://twitter.com/doctorow/status/982720929829675008
<whitequark> qu1j0t3: this is why FDA writes those thousand pages manuals yes
<whitequark> but you don't need to fix every kind of shenanigans.
<whitequark> you merely need to make it more expensive to adulterate than to provide genuine product
<whitequark> this is just like in infosec
<awygle> success! jeebus.
<whitequark> that took a while
<awygle> it's a process
<awygle> he's now on the powder _and_ a pill
<openfpga-github> [Glasgow-JTAG] awygle commented on issue #12: See https://github.com/KiCad/kicad-symbols/pull/461 for symbol PR https://github.com/whitequark/Glasgow-JTAG/issues/12#issuecomment-379516301
<awygle> kicad symbols - easier than drugging cats (TM)
<rqou> whitequark: afaik one reason china gave several of them the death penalty is because they made the chinese government lose face, and you _never_ want to do that
<rqou> random offtopic (drama?) question: who is @ErrataRob and why do people seem to dislike him?
GenTooMan has quit [Quit: Leaving]
<whitequark> rqou: famous infosec asshole
<rqou> ok then, why is he famous? :P
<whitequark> appears to be actually somewhat competent, unlike most famous assholes
<whitequark> well, i think he's famous for being a troll
<whitequark> he also did a few internet-wide surveys of exposed services (like vpn) with his "massscan" software
<whitequark> i'm not aware of any other notable achievements although they might exist
calle__ has joined ##openfpga
<awygle> he also seems to be in some kind of feud with the eff
<whitequark> don't read too much into it
<whitequark> he's constantly spouting aggravating nonsense because "trolling"
<whitequark> the SNR is just through the floor
noobineer has quit [Remote host closed the connection]
<whitequark> well that and just not generally caring about anything except his ass and popularity contests makes him pretty unpleasant
* whitequark shrugs
<whitequark> but that's, like, just my opinion
mumptai_ has quit [Ping timeout: 276 seconds]
<whitequark> awygle: wait, why are you modifying other level shifters
<awygle> i didn't on purpose, i think they auto-updated when i saved the library?
<whitequark> ah
<whitequark> `git checkout -p` the useless hunks maybe? it's just text
<whitequark> well, your call
<whitequark> i do it
<awygle> it's all the same data just in a different order. i'll leave it be unless somebody complains.
<whitequark> k
noobineer has joined ##openfpga
<whitequark> awygle: I know what to do with the three spare FPGA pins
<whitequark> put an eye-searing RGB LED there
<awygle> put it on the 400 mA driver pin
<whitequark> lol yeah what kind of FPGA does even *have* a 400 mA driver
<whitequark> idea
<whitequark> you could make a buck converter from it
<whitequark> anyway I mean these two pins are open-drain anyway, they're not *super* useful otherwise
<whitequark> three*
<whitequark> and if we also route I2C to the FPGA then we get two possibilities:
<whitequark> 1. polling ADCs/DACs on the FPGA
<whitequark> 2. connecting all the RGB LED hard IP to the *second* I2C core on FPGA to add nice indication
<whitequark> you could do weird shit like gradually shifting from red to green as the buffer fills :]
<awygle> whitequark: regarding https://github.com/whitequark/Glasgow-JTAG/issues/10 - am i looking at the wrong thing? this looks like it's just a SOT23-6
<whitequark> yes but why is it called TSOT-6
<awygle> SOT-23-6-THIN
<whitequark> oh.
<whitequark> THIN?
<whitequark> is it literally just low-profile?
<awygle> yup
<whitequark> lol okay
<whitequark> I guess the 3d model wouldn't match
<awygle> the kicad lib doesn't even have a 3d model currently
<awygle> not inaccurate by virtue of incomplete
<whitequark> excellent
<whitequark> I imagine you could add a package alias or something?
<openfpga-github> [Glasgow-JTAG] awygle commented on issue #10: This is just a SOT-23-6, except it's THIN. https://github.com/whitequark/Glasgow-JTAG/issues/10#issuecomment-379517722
<whitequark> since these are distinct packages after all, just with the same footprint
<awygle> yeah i think so
<awygle> although technically it's a _footprint_ lib :p
<whitequark> footprints are associated with 3d packages though, not devices, right?
<awygle> symbols have footprints have 3d packages
<awygle> oh you can give a footprint two 3d models
<awygle> i have no idea what the effect would be though
<whitequark> huh
<whitequark> any existing footprints using it?
<awygle> some generics
<awygle> maybe... one sec
<awygle> nope grep false positive. no footprint has >1 3d model.
<awygle> so bar some fiddly bits it looks like we have all the footprints now
<awygle> just a buncha symbols
<whitequark> yep
<whitequark> goddamn
<whitequark> the lattice devboard uses a $8.58 LDO
<awygle> wwwwwwwhy
<whitequark> guess which vendor made it
<awygle> i specced an LDO for $1.16 and felt vaguely like i was getting ripped off
<awygle> linear?
<whitequark> of course
<awygle> it better have like 10 nV RMS noise or something
<awygle> ... so i've been hearing a beeping noise all night. i just figured out what it is.
<awygle> the city installed those beeping crosswalk signals on the crosswalk directly under my windows.
<rqou> heh
<whitequark> 20 uV RMS
<rqou> at least it's more accessible this way?
<whitequark> it also tolerates ±20V on most pins
<awygle> rqou: i've never in my life complained to a city about anything but i might complain about this.
<rqou> why?
<rqou> too loud?
<awygle> too loud and never stops
<rqou> ah
<rqou> oh btw, they've been totally f*cking up hearst ave right now because they're building/improving the bus stop near euclid
<awygle> it's bad enough that they leave the (i'm speculating) 2500W light sculpture that's level with my window on until 2am every night
<rqou> this involved putting in new traffic lights
<rqou> but apparently the planners never considered that at intersections like this there are "strong incentives" to cross when the light is red
<rqou> such as "late for class"
<rqou> so the red lights are way too long, students cross anyways, and traffic is _worse_
<awygle> something something ban all cars
<rqou> something something transit is unusable
<awygle> yeah i nkow
<awygle> although if anywhere should ban all cars it's downtown berkeley
<sorear> if we replace all the cars with busses it won't be
<rqou> e.g. (quote stolen from @FioraAeterna) "there's a giant wall of darkness between san jose and fremont"
<sorear> i should spend some time there so I can actually compare whatever you have to what I've used
* awygle refuses to get drawn further into this discussion lol
<rqou> don't want to talk about the suckiness of BART?
<rqou> huh, there aren't too many actually good maps of _just_ how sucky transit is in the bay
<rqou> other than the "crash your pdf viewer" ac transit map
<azonenberg> awygle: multiple 3d models are stacked
<azonenberg> they're used for things like "to-220 with heatsink"
<azonenberg> or "dip-8 in socket"
<azonenberg> you can specify coordinates/scale independently for each
<awygle> argh
<rqou> wtf
<rqou> i guess that's useful
<rqou> btw azonenberg: you "develop" for kicad right? do you know whytf they still have "oldschool implicit ground 74xx" symbols?
<azonenberg> rqou: i'm not super active of a dev right now but i'm on the mailing list and send in patches periodically
<azonenberg> my guess? backward compat
<azonenberg> they dont want to remove stuff that breaks existing schematics
<azonenberg> Since current kicad has symbols stored by reference
<azonenberg> not embedded in the pcb like footprints
<rqou> oh wtf
<rqou> tell them to hurry up on eeschema-new :P
<whitequark> eeschema-new?
<rqou> eeschema rewritten with the new graphics layer
<whitequark> graphics layer?
<rqou> you know how pcbnew has three modes? the cairo/gl modes vs the legacy canvas mode
<whitequark> hrm
<rqou> for some reason legacy mode is slow as shit on macos
<rqou> which means that all of eeschema is laggy as shit
<rqou> (linux seems fine)
<openfpga-github> [Glasgow-JTAG] awygle commented on issue #9: See https://github.com/KiCad/kicad-symbols/pull/462 for Kicad PR https://github.com/whitequark/Glasgow-JTAG/issues/9#issuecomment-379519850
Bike has quit [Quit: Lost terminal]
rohitksingh has joined ##openfpga
rohitksingh has quit [Client Quit]
<azonenberg> rqou: i honestly dont understand why they have cairo
<azonenberg> like, are there seriously any computers left that don't run whatever opengl version they have as a minimum?
<rqou> i was told it's for when gl is busted for some reason
<rqou> yes
<rqou> the g3 potato imac i was testing :P
<rqou> yes, kicad does actually run on it
<azonenberg> IMO if you're trying to do EDA on a 15-year-old computer
<azonenberg> Go buy a new one :p
<rqou> but neither cairo nor GL work
<rqou> it actually works ok
<rqou> since the screen is also only like 800x600, pixel pushing isn't that slow :P
<whitequark> azonenberg: people seem to complain about missing opengl 2
<whitequark> which is absurd
<whitequark> then again people complain about dropping windows xp in solvespace
<rqou> wtf
<rqou> i would definitely drop that
<azonenberg> rqou: lol
<whitequark> i did
<rqou> the potato imac i was using actually runs a modern 4.x kernel
<rqou> so it's actually not nearly as unreasonable to support
<awygle> i wish cairo's' opengl backend wasn't essentially abandoned
<azonenberg> awygle: on that topic
<awygle> i would love a decent accelerated vector drawing library
<rqou> i really really want to see kicad-wasm
<azonenberg> libscopehal/scopeclient needs a GL port
<rqou> but it's so tainted with wx crap that that isn't really achievable
<azonenberg> for the UI
<azonenberg> its not a priority, but has to happen at some point
<rqou> i absolutely _hate_ wxwidgets
<azonenberg> rqou: it has a lot of windows-isms in it
<rqou> everything is "windowsy" and wrong and it encourages your program to become contaminated with it everywhere
<awygle> wx used to be great, like 10 years ago, and then seems to have just stopped
<azonenberg> compared to gtk
<azonenberg> and qt lives in their own little world where everything is qt :p
<awygle> wx was my first window toolkit or whatever we're calling those
<rqou> i heard gtk is a giant mess too with some diy C++ OO system
<whitequark> gtk is horrible
<awygle> gtk is awful. qt is actually pretty great.
<whitequark> and with gtk3 they dropped any premise of maintaining an ABI
<whitequark> or even a stable API
<whitequark> they just break shit in minor versions at will
<awygle> i seem to remember i had some license issue with pyqt though...
<rqou> this is why i've basically given up and decided that the "least bad" ui toolkit is called HTML5
<rqou> awygle: oh yeah that too
<azonenberg> gtk's C api is annoying fake-c++
<azonenberg> gtkmm hides that all behind actual C++
<awygle> i want a toolkit that maps some generic representation (HTML, CSS, XUL, idgaf what) into win32, cocoa, and either gtk or qt
<rqou> yeah, that's called a browser
<sorear> gtk4 at least has a policy of *which* minor versions they'll break shit at will in
<awygle> so that programs look like programs
<awygle> instead of marshmallows
<sorear> (yes, there is a gtk4)
<rqou> lolwut?
<azonenberg> awygle: i want a toolkit that maps some internal/generic representation into opengl :p
<rqou> re: marshmallows
<awygle> azonenberg: marshmallowssss
<rqou> also, can anyone explain to me why the gtk2->gtk3 migration was a Big Friggin Deal(TM)?
<awygle> rqou: GTK UIs, phone UIs, and web UIs all look like marshmallows
<rqou> how so?
<awygle> big rounded soft corners and tons of whitespace
<rqou> it doesn't have to look like that
<awygle> like a child's toy
<azonenberg> awygle: i *hate* marshmallow UIs
<azonenberg> everything designed for touchscreen looks awful to me
<whitequark> rqou: gtk3 is entirely gl-based
<awygle> azonenberg's UIs will end up looking like... what are those spikey things that fall from trees called?
<rqou> whitequark: wait, does it have a fallback?
<whitequark> azonenberg: behold, the new gedit https://twitter.com/whitequark/status/556588392092540928
<awygle> i predict this base slander solely on the basis of my reading of azonenberg's aura
<whitequark> rqou: yes. llvmpipe. lol
<rqou> why lol?
<rqou> does that not work?
<sorear> the older I get the more I miss libXaw
<whitequark> because the devices that need a gl fallback are precisely the same devices that can't run llvmpipe at more than 3 fps
<whitequark> netbooks mostly
<rqou> oh
<rqou> i assumed llvmpipe was for "didn't install the nvidia gl blob yet"
<whitequark> really old netbooks with that one really shitty intel GPU
<azonenberg> awygle: my dream UI style is kind of a hybrid of "windows 2000 style" and "windows 7 style"
<whitequark> mh
<rqou> poulsbo?
<rqou> is that _still_ a thing?
<azonenberg> rqou: poulsbo is where i'm moving in a few weeks
<awygle> isn't - yes
<rqou> lol
<azonenberg> was that an intel part codename?
<rqou> yes
<azonenberg> Lol, that's funny - i'll have to get my hands on one to decap
<rqou> the one where intel duct-taped a powervr sgx into their chipset
<whitequark> ^
<azonenberg> awygle: clean, minimalistic
<whitequark> it's
<rqou> with shitty-ass drivers
<whitequark> absolutely horrifying
<rqou> including no 64-bit support, even for windows
<whitequark> nouveau supports gl
<rqou> "supports"
<awygle> azonenberg: "jagged", "poorly anti-aliased" :p
<azonenberg> awygle: no frills, not the awful windows 98 style black/whie "drop shadows"
<whitequark> rqou: enough for gtk3 afaik
<azonenberg> modern antialiasing and maybe a tiny bit of rounding (like 3-5 pixel corner radius)
<rqou> someone please find a sploit in nvidia falcon secure boot plz
<azonenberg> but not marshmallow level
* awygle has switched from "kicad mode" to "shitposting mode"
<azonenberg> awygle: the Clearlooks gtk theme is the closest i've found to what i want
* awygle should really just crank out these last three symbols
<azonenberg> And i'm too lazy to go make my own one
<rqou> also someone please teach mesa/nouveau devs what an oscilloscope is so that reclocking can actually work
<whitequark> lol
<azonenberg> Clearlooks has a bit too much marshmallow for my liking still but it's not as bad as, say, win XP
<awygle> yeah xp was bad. i haven't had any issues with vista or later though. at least not from a "themes" standpoint
<awygle> although the new office products are extremely marshmallow
<azonenberg> awygle: win 8+ ruined things with the new start menu, navigating to anything became a pain
<azonenberg> and the run dialog either got hidden or removed
<azonenberg> everything is search now
<azonenberg> what if i actually just want to run devmgmt.msc?
<azonenberg> that was one of the things that drove me to linux
<rqou> you can type that straight into the search actually
<awygle> azonenberg: you can. literally just hit windows and type that.
<azonenberg> rqou: yes BUT it's a search
<awygle> or win+r if you really want "run"
<whitequark> azonenberg: win+r
<whitequark> the search is fucking slow though
<azonenberg> incidentally, recent linux distros are doing the same thing
<awygle> the search is slow if you don't know what you want
<awygle> if you do just type it and hit enter and don't wait for the suggestions
<azonenberg> with latest debian i havent figured out how to get the file open dialog
<whitequark> no, it can't keep up with my typing
<azonenberg> to not search
<whitequark> i type it and hit enter and it runs the wrong thing
<azonenberg> I'm used to the old style where i start typing a directory name and it highlights that dir
<awygle> huh, i've never had that problem
<rqou> oh yeah, debian/gnome/whoever did something "fun" recently and now nautilus is slow as shit
<azonenberg> now instead it trie to find everything with that in a substring
<azonenberg> including subdirectories
<azonenberg> annoying as heck
<rqou> oh azonenberg, someone already decapped poulsbo: https://en.wikichip.org/wiki/intel/chipsets/poulsbo
<azonenberg> rqou: yeah but it's not a 3x4 foot plot hung on my wall etc
<rqou> btw, just to make you jealous, <redacted> told me they had managed to acquire a (somewhat recent) unsawn intel wafer
<rqou> supposedly those are impossible to find
<rqou> oh wtf schwab added auto-enrolled shitty 2fa
<rqou> why do banks just _love_ doing shitty useless 2fa?
<rqou> one thing i just don't get is that doing shitty 2fa with sms or whatever costs _more money_ than TOTP
<rqou> so why not just use TOTP?
<awygle> why is this part number "DAC081C081C". what purpose is served by repeating the same suffix twice.
<whitequark> rqou: easier to lose an app than the SIM?
<whitequark> also if a SIM is taken out and replugged the bank can detect this
<rqou> they can?
<whitequark> yeah
<whitequark> my bank does
<rqou> hmm, i guess ss7 probably does have that info available
<whitequark> might vary by country/operator
<azonenberg> whitequark: is that just by detecting SIM power cycles?
<whitequark> azonenberg: no
<azonenberg> i.e. phone battery replacement
<whitequark> it looks for IMEI changes
<azonenberg> oh you mean moved to a new phone
<azonenberg> Not removed and reinserted
<whitequark> yes
<rqou> wait, that means that they _should_ in theory detect people doing ss7 hijacks
<rqou> that sounds like a huge pain in the ass though
<awygle> whitequark: my laptop's gonna die so i'm done for tonight
<awygle> i'll clean up the last two symbols tomorrow
<rqou> i guess i'm glad that in the US there's a unique state of *) consumers are basically never liable for fraud *) banks _suck_ at infosec
<whitequark> o/
* awygle should have spent at least 15-20 more minutes working instead of fucking around :p
<rqou> anyways, my experience with TOTP has been great
<rqou> i store all secrets on a yubikey neo with nfc
<rqou> (yes yes, it can be sniffed if you're close i guess)
<whitequark> i don't trust myself to not lose a hardware token i carry on myself
<rqou> i actually have _more_ problems with apps that decide that they're too special to tell you that they're actually TOTP (cough cough blizzard, valve)
<rqou> i just put it on my keyring
<whitequark> as it is, everything except server backups can be restored using just my head
<awygle> what whitequark said. i lose, like, everything
<whitequark> yeah i don't trust myself to not lose my keys
<whitequark> not being able to log into my house is mildly annoying
<whitequark> not being able to log into my mail is life-threatening
<awygle> also, "i've lost my keys" is not a time when i'd want to be locked out of all my accounts
<rqou> i do have a backup too on a flash drive
<whitequark> in your house?
<azonenberg> whitequark: my main file server is backed up to a raid array in a datacenter across the country
<rqou> yeah, and encrypted in the cloud
<whitequark> can you log into "the cloud" without TOTP?
<whitequark> azonenberg: i use tarsnap
<azonenberg> The SSH and PGP keys to access that (encrypted) backup are stored in a symmetrically-encrypted tar.gz.gpg
<rqou> hopefully you don't lose everything at once i guess
<azonenberg> That blob is stored on two identical thumbdrives in a bank vault
<azonenberg> The key to decrypt the blob lives only in my head
<whitequark> the key is printed on paper as datamatrix barcodes and laminated
<rqou> i have yet to find a good solution for "lost absolutely everything"
noobineer has quit [Remote host closed the connection]
bitd has joined ##openfpga
<azonenberg> rqou: My system is robust to loss of one site (home, bank, or datacenter)
<azonenberg> plus failure of any one storage device in each of the surviving two sites
<azonenberg> e.g. one hdd and one thumbdrive
<azonenberg> That's sufficient redundancy for me
<whitequark> i guess it won't survive a fire but it'll survive flood and mold
<rqou> i need to actually plan out exactly how much redundancy i do have
<whitequark> so that's fine by me
<bitd> But what about redundancy of the redundant system >.>
<azonenberg> If both me and my wife somehow lose our keys to the safety deposit box, after proving our identity to the bank we can get the box drilled out if worst comes to worst
<bitd> Morning >.>
<azonenberg> rqou: This is full N+1 on everything, afaik
<awygle> i have very little redundancy in my system generally but my life is very privileged so it's not a big deal
<rqou> i have the key to decrypt the backup flash drive split with shamir's algorithm and i have some friends keep some of the other key shares
<azonenberg> i havent found a single point of failure yet
<azonenberg> Other than my brain, but if i die / am brain damaged enough to not remember the 128-bit hex value I used as a password
<azonenberg> That data isn't much good to me anyway :p
<rqou> also, tools for using secret sharding suck
<rqou> i guess that's not a particularly common operation?
<bitd> rqou, thats a lot of trust you are putting in your friends.
<rqou> they don't have all the shares
<azonenberg> rqou: if you want to do a simple N+1 sharing scheme
<whitequark> ... so effectively you've only reduced *your* capacity to decrypt the thumbdrive?
<azonenberg> you can basically do raid5
<azonenberg> raid6*
<rqou> er no, because i keep multiple keyshares
<rqou> i'm probably going to fix this at some point though
<rqou> you can apply secret sharing multiple times to get a hierarchical system of keys
<bitd> Just seems like having to rely on other people when shit hits the fan... wouldnt put me at ease.
<rqou> yeah well i'm not great with memorized secrets
<rqou> especially ones that won't get used very often
<bitd> Use mnemonics.
<azonenberg> rqou: I dont memorize a lot of secrets
<bitd> You can grind it in, in one go.
<azonenberg> I memorize a small handful then use them to access other stuff
<azonenberg> one 32-char hex plus a handful of 8-char alnums
<rqou> yeah well you're brain is overclocked :P
<rqou> *your
<rqou> herp
<azonenberg> Everything that i use less frequently lives in a db that i can read them out of as needed
<azonenberg> the most frequently used entries, like say my online banking password, eventually end up getting memorized b/c i type them in so often
<bitd> Probably situation dependent memory though.
<rqou> hmm yeah, i still have a legacy truecrypt volume for that
<rqou> i need to do another round of cleanup soon
<bitd> Can you type it in when at a different location?
<azonenberg> bitd: yes
<azonenberg> The bulk of the passwords in that db, like say my facebook, i dont have access to when not home
<azonenberg> I almost never use that one, i keep it logged in 24/7 in the social media VM
<azonenberg> So i never have to log in
<azonenberg> As with my other "general purpose" passwords it's 16 char alnum
<azonenberg> The only ones i memorize are those that are either important enough i dont want to leave it logged in / the system doesn't let me (bank)
<azonenberg> or those that i use for local console login (work laptop, home desktop root, etc)
<azonenberg> everything else i just log in to decrypt things then have the dump where i need it
<bitd> I use a different 32 char password for everything.
<azonenberg> bitd: I have a few 8 chars that i reuse all over the place on forums etc that have no real value to an attacker
<azonenberg> As soon as an account gains any actual value i switch to a dedicated 16 alnum i don't use anywhere else
<azonenberg> log2(62^16) is about 95 bits of entropy
<azonenberg> Nobody is bruteforcing that
<rqou> at this point for security i mostly rely on TOTP and don't place particular value on passwords
<azonenberg> and if the system keeps it in cleartext, IDGAF
<rqou> (effectively downgrading to 1 factor)
<rqou> i just dump all passwords in mozilla's password manager
<azonenberg> they couldn't have gotten that password without pwning the server, at which point my password is compromised anyway
<bitd> Well I use the object I want access to in my mnemonic to unlock the story that contains the 32 chars :)
<bitd> I have no log of any password, ever.
<rqou> passwords are pretty awful and pointless
<azonenberg> bitd: I dont really care about most of them
<azonenberg> email and banking are the handful that really matter to me
<rqou> azonenberg: what would the potential problems be with doing TOTP/U2F only, without passwords?
<azonenberg> rqou: Dont know, never really thought about it
<rqou> heh, i've given up on banking
<azonenberg> a lot of stuff i support with doesn't support it
<rqou> us banking security is a joke
<azonenberg> so i havent had time to look into it
* azonenberg got his CC skimmed a few weeks back... you don't have to tell me that
<bitd> Uf..
<azonenberg> At a gas pump, about the only place around here that is still using magstripes :p
<rqou> and consumers aren't liable, so whatever
<rqou> lol bullshit
<rqou> i still use magstripe all the time
<azonenberg> I only use my physical CC at walmart, the grocery store, and the home depot
<azonenberg> plus gas
<azonenberg> pretty much
<whitequark> lol magstripe
<azonenberg> oh, and sometimes lunch at work
<whitequark> i use... contactless visa
<azonenberg> everything else is online
<rqou> whitequark: lol you're funny
<rqou> contactless in the us is also a joke
<whitequark> i used to hate it but hk's octopus card was just *too* convenient
<rqou> although it actually did work the few times i tried it
<whitequark> however, the cash advance fees are not
<azonenberg> whitequark: when i was in HK i used octopus for probably 80% of my transactions
<rqou> yeah octopus is great
<whitequark> yeah i used to do that
<bitd> They introduced contactless payments here a while ago. Now a lot of 'people' are going around with bags containing mobile pin equipment. If they are holding a bag against your leg, its not because they like you/
<azonenberg> on that note, how long is an unused octopus good for?
<rqou> afaik forever?
<whitequark> yeah
<rqou> my card is from more than a decade ago
<azonenberg> they dont disable after a year or so of inactivity?
<whitequark> the problem with octopus is you cant hold over 1000 HKD on it
<azonenberg> as in, i havent been in HK since december 2016
<whitequark> and you have to withdraw cash
<azonenberg> is mine still any good?
<whitequark> azonenberg: definitely
<whitequark> there's no real point in expiring them
<whitequark> they're phasing out some extremely old cards
<rqou> they are?
<azonenberg> rqou / whitequark: well seeing as i likely won't be going back to HK any time soon
pie_ has quit [Ping timeout: 260 seconds]
<whitequark> yeah, there are automatic replacement machines in MTR
<rqou> did someone find a vuln in felica?
<azonenberg> If either of you is in the seattle area in the near future
<azonenberg> You can have mine
<azonenberg> along with whatever balance is on it
<azonenberg> probably only a few tens of HKD, but seems a shame to let it go to waste
<rqou> azonenberg: i dare you to find a vuln in felica
<whitequark> decap it lol
<rqou> anyways, i wonder if the card that i have needs replacement
<rqou> it's definitely 10 years old
<whitequark> how does the serial number look like
<azonenberg> whitequark: i actually have an annoying problem as a result of all of my various international travel from work
<whitequark> if it doesn't have (N) it is too old
<whitequark> otherwise keep it
<azonenberg> I have a small amount (probably <50 USD equivalent of each, possibly <10) of many foreign currencies
<azonenberg> not enoguh to be very useful when i go to that country
<rqou> whitequark: it ends in (0)
<azonenberg> and small enough that if i take it to a bank the exchange fees will eat most of the value :p
<whitequark> azonenberg: i usually exchange them at the airport at the draconian rate they offer
<azonenberg> whitequark: the exchange place in SEA charges like a 10 USD conversion fee PLUS an awful rate
<azonenberg> i'd have literally nothing left after that
<rqou> whitequark: are they already rejecting old cards?
<whitequark> rqou: not sure
<whitequark> but (0) is new enough
<rqou> ah ok
<azonenberg> my local bank branch has much better rates but it's still a lot of effort for... not even enough to buy lunch
<whitequark> yeah
<azonenberg> right now they just sit in a pile at the back of my safe gathering dust
<whitequark> at some point i just tossed out all coins i had
<azonenberg> i think i have CAD, HKD, GBP, and EUR right now
<whitequark> not worth their weight
<whitequark> bills are easy to keep
<bitd> Ugh. back to preping my thesis defence :(
<azonenberg> whitequark: yeah right now i have everything
<azonenberg> in ziplocs
<azonenberg> the coins are nice paperweights on top of the envelopes holding the bills
<rqou> btw if anybody wants to test all sorts of NFC bullshit: https://photos.app.goo.gl/7Nst7i4r3mATgmrf1
<azonenberg> actually more like bill, i probably only have one of each :p
<rqou> i strongly advise not altering the data though, you wouldn't want to be asked to "have a chat" with the PSB or anything :P
<whitequark> PSB?
<rqou> public security bureau (mainland)
<whitequark> ah
<rqou> i might also have a SuiCa card lying around somewhere (which is also felica)
<rqou> btw that yellow card in the photo is mifare classic lol
<awygle> I currently don't even have the currency of my own country
<awygle> I should get some emergency cash together again at some point
<rqou> oh yeah, btw if anybody has sf bay Clipper card encryption keys, i have a "special" card that might be worth investigating
<azonenberg> awygle: yeah i like to have at least enough to pay for taxi fare if my bike breaks down
<azonenberg> never know when you might find the one cab without a working CC reader :p
rohitksingh has joined ##openfpga
rohitksingh has quit [Client Quit]
rohitksingh has joined ##openfpga
<whitequark> hm
<whitequark> is there any way to make usbmon work "not like shit"?
rohitksingh has quit [Quit: Leaving.]
<rqou> lol
<rqou> i usually capture with Wireshark
<whitequark> yes
<whitequark> it doesn't work properly
<whitequark> URBs are truncated
<rqou> hmm, never saw that before
<whitequark> and I don't see completions until the next request arrives
<rqou> ping digshadow?
<rqou> also, try using sudo? :P
<rqou> whatever the "thing" that debian did to make Ethernet capture not require sudo doesn't seem to work for usb
<azonenberg> rqou: they made the dumpcap binary setuid root
<azonenberg> then the wireshark binary talks to it
<azonenberg> This is a non-default debconf option
<rqou> unless you're debugging fx2 firmware, in which case I'm surprised you haven't panicked the kernel yet :P
<rqou> azonenberg: oh what
<rqou> that sounds like a new security hole waiting to happen
<rqou> see: beep :P
<azonenberg> no its a net gain - capturing packets always needs root
<azonenberg> this means the UI isnt root
<azonenberg> where all the parsing is
<azonenberg> only the actual capturing
<rqou> cap_net_admin? :P
<rqou> (which i heard was historically basically equivalent to root)
<whitequark> no
<whitequark> i chown'ed the entire usbmon tree to my user
<whitequark> oh hm
scrts has quit [Ping timeout: 265 seconds]
<whitequark> ohhh
<whitequark> I needed to fix the permissions on /dev/usbmon
<whitequark> not on /sys/kernel/debug/usb/usbmon, which is legacy shit
<rqou> huh i didn't know that either
<rqou> also, why does it appear to me that linux backwards compatibility isn't nearly as insane as windows backwards compatibility?
<rqou> e.g
<rqou> "we just kept some extra files around" vs "the stack frame layout for this callback needs to be manually constructed in assembly"
<azonenberg> rqou: because windows expects to work with binary blobs from decades ago
<azonenberg> linux doesnt care as much about that as you can normally recompile things from source if there's an API-compatible ABI change
<rqou> er, many binary blobs from decades ago still work on linux
<azonenberg> not in the kernel they dont
<rqou> you might have to reenable vsyscall
<azonenberg> kernel doesnt have abi compatibility whatsoever afaik between versions
<rqou> sure
<whitequark> most blobs aren't in kernel
<rqou> ^
<rqou> although i guess linux backwards compatibility occasionally involves a "mauro shut the **** up"
<azonenberg> also, windows expects you to use their one and only userland api
<azonenberg> the kernel syscall interface is not stable
<azonenberg> linux has the public api be the syscall interface
<azonenberg> and libraries can change freely
<rqou> hmm yeah, i think that helps a lot
<rqou> especially since userspace f***s itself up all the time
scrts has joined ##openfpga
<rqou> e.g. whatever poettering and/or fd.o keep doing
<rqou> btw random question: are the specific values of the linux 32/64-bit user mode code segment selector considered part of the kernel abi?
<rqou> you can make some really really fun frankenprocesses this way
<rqou> whereas apparently win10 explicitly did something to make this not possible
<rqou> oh yeah, azonenberg how would a 32/64-bit mixed frankenprocess make you feel? :P
<azonenberg> rqou: ...
<rqou> what? :P
<rqou> you can use either type of syscall from either type of process
<rqou> unlike windows where you _have_ to go through the ntdll thunks
<azonenberg> i meant more like, changing from protected to long mode and back
<azonenberg> in the middle of one app
<rqou> why not?
<rqou> apparently it's even better on windows because it'll make attached debuggers lose control
<rqou> idk if gdb handles this correctly
<whitequark> rqou: do you just do long calls to cs:eip/rip?
<whitequark> that actually sounds disturbingly useful
<rqou> iirc yes
<whitequark> you could load 64-bit DSOs into 32-bit processes
<rqou> yes
<whitequark> with thunking
<rqou> er, maybe
<rqou> you might have to mess with prctl
<whitequark> hm
<rqou> which isn't a privileged syscall, so you can just do that
<whitequark> rqou: also what do you have against debugging fx2
<openfpga-github> libfx2/master e0e7f70 whitequark: Use English (US) in the langid descriptor....
<openfpga-github> libfx2/master f8e82ba whitequark: Unbreak rebuilding of libfx2 library as a dependency after cleaning.
<openfpga-github> [libfx2] whitequark pushed 2 new commits to master: https://github.com/whitequark/libfx2/compare/db5fc374c264...e0e7f708f9dd
<rqou> nah, that was just a jab at how buggy usb driver stacks have historically been
<whitequark> ah
<whitequark> no, linux has been remarkably stable so far
<rqou> e.g. the facedancer person causing X/Skype, etc. to crash
<rqou> iirc these both had format string vulns
<rqou> also the ps3 gameos
<whitequark> lol
<rqou> which caused a nice temporary shortage of usb atmegas :P
* whitequark stares at code
<whitequark> so, bitstream download was failing.
<rqou> also, afaik the syscall you might have to poke to load a 64-bit dso in a 32-bit process is apparently personality(2) not prctl
<whitequark> for a reason i couldn't understand.
<whitequark> turns out it was failing a check i put there to avoid a bug where i would forget to increment an index, because i forgot to increment that index
<whitequark> defensive programming : bugs 1 : 0
<rqou> nice
<rqou> also: "PER_HPUX (since Linux 2.4) Support for 32-bit HP/UX. This support was never complete, and was dropped so that since Linux 4.0, this value has no effect."
<rqou> brilliant
<rqou> "PER_LINUX32 (since Linux 2.2) [To be documented.]"
<whitequark> ...
<rqou> i get the feeling nobody really touches low-level syscalls like this very often
<whitequark> i wonder why
<rqou> apparently also according to rich felker the reason the linux mips abi is so weird and gratuitously different is because they were trying to pretend to be an irix-like
<whitequark> yes
<whitequark> linux mips is horrifying
pie_ has joined ##openfpga
pie__ has joined ##openfpga
pie_ has quit [Ping timeout: 260 seconds]
Lord_Nightmare has quit [Quit: ZNC - http://znc.in]
Lord_Nightmare has joined ##openfpga
Lord_Nightmare2 has joined ##openfpga
Lord_Nightmare has quit [Ping timeout: 240 seconds]
Lord_Nig- has joined ##openfpga
Lord_Nightmare2 has quit [Ping timeout: 260 seconds]
Lord_Nightmare has joined ##openfpga
Lord_Nig- has quit [Ping timeout: 264 seconds]
<whitequark> ugh ISRs are multithreading
<rqou> no, even worse
<rqou> they don't get blocked by mutexes
<whitequark> I don't have mutexes lol
<whitequark> just spent some hours figuring out race conditions between USB setup interrupt code and main loop
<whitequark> ended up just using a semaphore because anything more complex kept falling apart
<whitequark> if you ever write 'volatile' look good and hard in the mirror and ask yourself where the bug is
<whitequark> tbh i have newfound appreciation for japaric's thing
<rqou> it's pretty neat
<rqou> the hals are still quite lacking
<rqou> SVDs are still a clusterfuck
<whitequark> ehh SVDs are fine by me
<whitequark> well the ones I maintain :P
<rqou> not if you don't want to duplicate effort everywhere
<rqou> wait, you maintain some?
<whitequark> dslite2svd
<rqou> are we duplicating effort?
<whitequark> ti tm4c129x
<rqou> dslite?
<whitequark> ti doesn't provide svd files
<rqou> oh, this is some TI thing? not some nintendo thing? name collision lol
<whitequark> they provide some sorta fuckery for their IDE that's basically the same thing as SVD, except it's a different XML format
<whitequark> so I wrote a converter from dslite to svd
* whitequark . o O ( wonder if I could've done it via xslt )
<rqou> i've been trying (important: trying, not succeeding) to maintain a set of unified stm32 SVDs
<rqou> as well as EFM32
<rqou> EFM32 SVDs are ok, STM32 is a clusterfuck
<rqou> also, you can't generate device crates that support multiple devices
<rqou> nor can HALs support multiple devices
<rqou> duplicated effort, duplicated effort everywhere
<whitequark> that doesn't seem to bother you in case of kicad libs
<rqou> it kinda does, but i don't think your approach fixes anything
<rqou> i would really like to write a parametric footprint tool though
<rqou> i would be totally on board with that
<whitequark> meh
<whitequark> niche case
<rqou> idk, there have been many cases where i decided not to use an existing lib because i wasn't happy with some parameters
<openfpga-github> [libfx2] whitequark pushed 2 new commits to master: https://github.com/whitequark/libfx2/compare/e0e7f708f9dd...e62d8776c534
<openfpga-github> libfx2/master e62d877 whitequark: Eliminate race conditions in SETUP request handling....
<openfpga-github> libfx2/master 53dac29 whitequark: Fix docs after e99a375e.
<rqou> but overall i would summarize it as "i can see hope in the state of software improving, but not the state of EDA libraries improving"
<rqou> maybe/hopefully you'll prove me wrong
<rqou> whitequark: after you get your board back, maybe you should really quickly comment on *) how many footprint errors there were and how severe *) how many of your kicad lib PRs got merged
<rqou> whitequark: random question: is there a somewhat-common-ish idea of how a usb _device_ framework should work? (i.e. the logical counterpart to libusb)
<rqou> or does everybody have their own ad-hoc thing?
<whitequark> I'm not familiar
<whitequark> I was always stuck in cases where only the ad-hoc thing worked
<whitequark> e.g. avrusb
<rqou> lol
<whitequark> what
<whitequark> it was great!!
<rqou> embedded sucks, news at 11 :P
<rqou> isn't avrusb the proprietary one?
<rqou> oh that
<rqou> that is pretty neat
<rqou> i always saw it referred to as V-USB, so i was confused
<whitequark> it used to be called avr-usb
<whitequark> very very long ago
<whitequark> around 2007, maybe earlier
<rqou> ah ok
<whitequark> damn i'm old
<rqou> i didn't use it until ~2009 or so
<rqou> anyways, that also means that AVRs have three usb stacks: V-USB, atmel's, and LUFA
<rqou> and they're all different, because reasons!
<whitequark> LUFA?
<whitequark> oh, myusb
<rqou> LUFA is for the AVRs with actual USB hardware
<rqou> written by some guy in australia
<whitequark> hooooooly shiiiit
<rqou> and then ST has their own USB stack, and silabs has their own, and TI has their own, and NXP has their own, etc.
<whitequark> my code for downloading bitstream to FPGA worked the first time
<whitequark> i ran it
<rqou> nice
<whitequark> my retinas ._.
<rqou> whitequark: dumb question: what exactly is xargo and why is it magic?
<rqou> why is cargo not sufficient to do whatever it is that xargo actually needs to do?
<rqou> why is this part of rust not documented at all?
<whitequark> cargo can compile crates that are, say, in crates.io, or a repository, or something like that
<whitequark> libcore and libstd are a part of rust and they're implicitly expected to be found by rustc in its sysroot
<whitequark> xargo takes the rust source distribution that comes with rustup, pulls the necessary libs out of it, builds them, and puts them in a sysroot
<whitequark> that's more or less it
<whitequark> at M-Labs we don't use xargo because it sucks if you have a fork of rust
<rqou> wait why do you have a fork?
<whitequark> instead I have a 10-line shell script that builds a sysroot for our architecture and that's... sufficient
<whitequark> uh
<whitequark> because OR1K isn't in upstream LLVM and so it can't be in upstream Rust
<rqou> ugh
<rqou> that's dumb
<rqou> wait, it isn't?
<whitequark> it isn't
<rqou> i swear i saw plumbing for or1k support
<whitequark> no one has upstreamed it
<whitequark> doubt it
<whitequark> it has never been there afaik
<rqou> hmm
<rqou> maybe just in libunwind
<rqou> so why isn't it upstream yet?
<whitequark> oh
<whitequark> yea, I upstreamed the libunwind patches
<whitequark> because those don't really depend on anything
<whitequark> why? because upstreaming a backend into LLVM is a multi-month commitment and then you have to maintain buildbots forever
<whitequark> I mean, it makes sense, you want a backend to not be shit and you don't want it to break
<whitequark> but M-Labs isn't gonna let me do that
<rqou> and now everybody ends up with their own special snowflake llvms
<whitequark> that's better than an unmaintained piece of shit upstream
<whitequark> when you use "some random person's llvm" you have the right expectations
<rqou> so rust can't use an external llvm?
<whitequark> sure it can
<whitequark> ./configure --llvm-root
<rqou> then why do you need a special fork?
<whitequark> because rust needs to understand the platform C ABI
<rqou> i thought there was a magic file that told it that?
<whitequark> and--now *this* is very stupid--llvm doesn't include an implementation of C ABI
<rqou> (except for the part where all such examples seem to be bitrotted and don't work anymore)
<whitequark> so everyone rolls their own shitty C ABI
<whitequark> no, that's different
<whitequark> that file is like the gcc specs thing
<whitequark> it tells rustc what kinda files to write, how to invoke the linker, which cfg() switches to activat
<whitequark> however C ABI is defined via code only
<rqou> but supposedly it also has things like "The LLVM data layout to use."
<whitequark> that one is actually really dumb
<whitequark> LLVM targets know their data layout perfectly well
<whitequark> I'm not sure why rustc even needs that field
<rqou> idk, none of the examples of rust target files seem to work anymore :P
<whitequark> no, it does need it
<whitequark> but I don't know why
<rqou> everybody who isn't doing cortex-m seems to just shrug and give up
<whitequark> because they're weak?
<whitequark> seriously, you simply go and read the source of rustc_trans
<whitequark> it's extremely well documented
<whitequark> and you have several dozens of examples right inside of it
<rqou> hmm, so "cpu" and "data-layout" in the target file are both not sufficient?
<rqou> then what's the point of target files?
<whitequark> uh, no
<whitequark> let's say you have ARM. the entire ARM platform uses a few defined ABIs.
<whitequark> well, nothing except EABI matters anymore, so basically one ABI
<whitequark> however it has a bazillion possible CPUs
<whitequark> and even more target features
<whitequark> the only case where you can't use a target file is when you need a new LLVM backend
<whitequark> well, there's also the stupid reason for needing to fork rust, namely the fact that you need to call LLVMInitializeOR1K{AsmParser,AsmPrinter,...}
<whitequark> but that could probably be automated into a build script if someone really wanted it
<rqou> ah, so target files would be used for e.g. armv4shittycoprocessoredition-vendorros-eabi
<whitequark> yes
<whitequark> that's basically all they're useful for
<whitequark> well, also things like overriding a linker
<rqou> i thought cargo/xargo can just override the linker?
<whitequark> how do you think it does that
<whitequark> well
<rqou> oh
<rqou> lol
<whitequark> you can do it via [target.$triple] section in .cargo/config
<whitequark> but I think xargo does it via target files
<whitequark> oh and you could also do RUSTFLAGS=-Clinker
<rqou> huh, there's a whole "guide to rustc development" now
<rqou> i swear that wasn't here before
<whitequark> yes
<whitequark> it's a recent addition
<whitequark> and a good one, it's a royal pita to develop rustc if you don't have obscure domain knowledge
<whitequark> primarily because, unless you know exactly how to invoke its build system, EVERYTHING takes an eternity
<rqou> yeah, last i looked there were just random blog posts telling you to pass a bajillion flags to "./x.py"
<whitequark> it's like building LLVM, if LLVM had ten compile units instead of two thousand
* whitequark shudders
<rqou> LLVM builds in a reasonable amount of time for me lol
<whitequark> doesn't help that the fucking thing really wants to bootstrap itself at least twice
<whitequark> well
<rqou> i also have 20 threads
<rqou> :P
<whitequark> my laptop builds LLVM in something like 40 minutes for a release build from scratch
<whitequark> and maybe 25 for a debug one
<whitequark> that's with gold, -gsplit-dwarf and ninja
<whitequark> hm
<whitequark> I cna download bitstream in around 800ms
<whitequark> should I optimize it further
<rqou> meh
<rqou> good enough :P
<whitequark> Why The Hell Not
<rqou> lol the entire "Generating LLVM IR" section of the rustc guide hasn't been written yet
<rqou> i guess this part must be black magic :P
<whitequark> not really
<whitequark> that's one of the simplest parts
Bike has joined ##openfpga
<whitequark> ... oh
<whitequark> lol
<whitequark> i optimized the part of it that wasn't slow
<rqou> i guess i'm glad i didn't try to add ppc64 musl support to rust
<whitequark> oh?
<rqou> (not that i ever got ppc32 musl working either)
<whitequark> how is that wrong?
<rqou> musl only supports ELFv2 regardless of endian
<whitequark> huh]
<rqou> iirc rich felker thinks ELFv1 is dumb :P
<rqou> ah "has function descriptors, so you cant just reference a function pointer, mainly makes asm code messier as needs a header for every function."
<rqou> "v2, usually known as powerpc64le, helpfully. It is not little endian exclusively, but there are no existing distros that are big endian."
<rqou> but afaik this distinction is a userspace problem and the syscall interface works the same way
<rqou> why are all linux ABIs a disaster in some way?
oeuf has joined ##openfpga
digshadow has joined ##openfpga
egg|egg has quit [Ping timeout: 264 seconds]
eduardo_ has joined ##openfpga
eduardo__ has quit [Ping timeout: 260 seconds]
<whitequark> wtf why do lattice technotes recommend obviously wrong things
<rqou> Vendors(TM)
<whitequark> it says use 49 dummy SCK cycles in one place and 100 in another
<whitequark> and the real number that the FPGA uses is 14
<rqou> tololo
<rqou> *trololo
<rqou> damn it's late
<whitequark> this can lead to bus contention...
* rqou zzzz
<whitequark> well I guess if a 8mA driver is fighting with a 4mA one
<whitequark> not much bad can happen
<openfpga-github> [Glasgow-JTAG] whitequark pushed 1 new commit to master: https://github.com/whitequark/Glasgow-JTAG/commit/6d2b0d4cfb4bfe6379390acd365a2b3593cd06c2
<openfpga-github> Glasgow-JTAG/master 6d2b0d4 whitequark: Implement FPGA bitstream downloading.
pie__ has quit [Quit: Leaving]
soylentyellow has quit [Ping timeout: 256 seconds]
soylentyellow has joined ##openfpga
rohitksingh has joined ##openfpga
calle__ has quit [Remote host closed the connection]
soylentyellow has quit [Ping timeout: 240 seconds]
soylentyellow has joined ##openfpga
pie_ has joined ##openfpga
rohitksingh has quit [Ping timeout: 255 seconds]
soylentyellow has quit [Ping timeout: 276 seconds]
wpwrak has quit [Ping timeout: 240 seconds]
soylentyellow has joined ##openfpga
wpwrak has joined ##openfpga
rohitksingh has joined ##openfpga
<awygle> good morning world
<Zorix> shh.. we are still sleeping
noobineer has joined ##openfpga
* qu1j0t3 snores, rolls over
mumptai has joined ##openfpga
pie_ has quit [Ping timeout: 264 seconds]
rohitksingh has quit [Quit: Leaving.]
pie_ has joined ##openfpga
rohitksingh has joined ##openfpga
X-Scale has quit [Ping timeout: 264 seconds]
rohitksingh has quit [Quit: Leaving.]
GenTooMan has joined ##openfpga
rohitksingh has joined ##openfpga
<kc8apf> makerchip ide seems to be a fancy demo for TL-Verilog and nothing more
X-Scale has joined ##openfpga
<mithro> Morning
<kc8apf> *yawn*
wpwrak has quit [Ping timeout: 240 seconds]
<cr1901_modern> Morning? But it's the afternoon.
rohitksingh has quit [Quit: Leaving.]
X-Scale has quit [Ping timeout: 256 seconds]
wpwrak has joined ##openfpga
<awygle> as always, in the pacific northwest it's Grey O Clock
<kc8apf> awygle: I've seen sun there once
<awygle> it's actually lovely in the summer, we just don't tell californians that :p
<awygle> today is fulfilling every stereotype though
<kc8apf> they figured it out. Hence all the techies moving there
<awygle> "SAD" vs "Bay Area housing prices" is not as easy a choice as it once was i guess
<kc8apf> prices are getting pretty close recently
<kc8apf> Kirkland/Bellvue seem to be on par with San Jose
rohitksingh has joined ##openfpga
<pie_> HM
<pie_> *hm
<pie_> wait, last release 2010? :/
<mithro> I also came across this recently -> https://github.com/CircuitData/CircuitData-Language
rohitksingh has quit [Quit: Leaving.]
<awygle> rqou should use that to implement his parametric footprint generator
m_w has joined ##openfpga
<kc8apf> mithro: CircuitData-Language seems in the vein of IPC-2581. Many attempts to standardize PCB manufacturing data formats but no real success.
<awygle> which is a bummer
<awygle> whitequark: every component is now in kicad PR form
<awygle> i have other stuff to pivot to for probably the rest of today but i'll keep an eye on the PRs and address anything that comes up from the maintainers
<awygle> Escaping BGAs - always keep one hand on the left wall
<azonenberg> awygle: lol
<azonenberg> brings back memories of the fire tower
<azonenberg> crawling through in pitch darkness with one hand on the ankle of your buddy and the other on the wall
X-Scale has joined ##openfpga
pie__ has joined ##openfpga
pie_ has quit [Ping timeout: 276 seconds]
<pie__> oh god i just figured out ive been doing the signs on subdeterminant expansion wrong for a while now...
<pie__> bleh, *expansion by minors
<sorear> determinants are weirdly explained in a lot of places
<pie__> i was doing the shape of things right but forgot the parity base sign change :/
<pie__> (just 3x3 stuff though)
<pie__> *parity based
bitd has quit [Quit: Leaving]
<pointfree> 17:29 <awygle> as always, in the pacific northwest it's Grey O Clock
<pointfree> awygle just infected me with a weird thought about a GrAy-code clock. https://www.mathworks.com/matlabcentral/mlc-downloads/downloads/submissions/40928/versions/3/screenshot.jpg ...I'd buy a Gray code wristwatch.
<pointfree> ...maybe that could actually be useful in some way: easily subdivided endlessly, combine with sequential logic and logic minimization to minimize expressions with relative timing.
<pointfree> The direction of the microwave plate reverses from the previous direction every time I turn it on again.
<pie__> is...the microwave plate related to the gray code
<pie__> because yeah i thin ka lot of microwaves do that
<pointfree> pie__: it most likely uses a gray-code rotary encoder.
<pie__> huh.
<pie__> why would a microwave need to keep track of plate rotation
<qu1j0t3> oh, yeah, why aren't grey coded DACs more common
<qu1j0t3> gray*
<pie__> *rainbow
<pointfree> each digit of a gray code would be a frequency. Mask digits to filter out frequencies
mumptai has quit [Quit: Verlassend]
pie__ has quit [Ping timeout: 255 seconds]
pie_ has joined ##openfpga
<awygle> ugh pcb assembly is the worst lol
pie_ has quit [Ping timeout: 264 seconds]
<Prf_Jakob> Tonights progress on the crazy plan of using DDR3 SODIMM as a FPGA breakout board. https://i.imgur.com/q6uPzFK.png
marcan has quit [Ping timeout: 240 seconds]
marcan has joined ##openfpga
<awygle> ooo so many dimensions
<Prf_Jakob> Heh, thanks. Kicad's learning curve is not the friendliest.
<awygle> very true
<awygle> well that's cool, icecube thinks my host id doesn't match even though it does. great.
<awygle> why the hell do they even have licenses on a tool that is 100% free