<mithro>
Batavia is an implementation of the Python virtual machine, written in JavaScript. With Batavia, you can run Python bytecode in your browser.
<rqou>
ah, that's more likely to work correctly
<rqou>
but yeah, there are tons of python->JS transpilers
<mithro>
rqou: They have a pretty big test suite
<rqou>
it is/was a pretty popular "final project" at one point
<rqou>
yeah, reimplementing a python VM might actually wor
<rqou>
*work
<awygle>
hm it appears that qt uses widgets that look native but are not
<rqou>
usually i see transpilers that take a lot of shortcuts and fail when you use e.g. metaclasses
<mithro>
awygle: Correct - they are frustratingly close and fall into the uncanny valley
<rqou>
awygle: can you explain to me why this really matters?
<awygle>
rqou: mithro just said it, the uncanny valley. non native apps feel wrong, they don't integrate well with the system, it's just a bad user experience
<rqou>
hmm, i guess i just don't expect things to work this way
<awygle>
rqou: you don't feel this way because you change platforms a lot and your primary platform is Linux which has never had a platform UI toolkit
<mithro>
awygle: The thing is you don't really expect a web app to integrate with the system properly, but something which kind of does look like it should you then get frustrated when things don't work
<awygle>
so all Linux gui apps are frustratingly terrible
<awygle>
Well not all but lots
<awygle>
And yes browsers take the R2-D2 approach
<mithro>
R2-D2 approach?
<awygle>
Cute and friendly and not even trying to look human
<awygle>
The other side of the uncanny valley
<rqou>
what about "vidya" custom UIs? :P
<rqou>
somehow they all seem to have one
<mithro>
So, anyone want to review some schematics for me?
<rqou>
isn't the oval discussion basically a silly bikeshed?
<rqou>
(i have no opinion either way and don't usually use rounded pads)
<whitequark>
i dunno, i find the name confusing, and awygle does too
<rqou>
also, f*cking sigrok
<rqou>
why does it crash all the time?
* rqou
blames USB
<whitequark>
are you running trunk?
<rqou>
no
<rqou>
whatever is in debian sid
<whitequark>
oh
<whitequark>
well debian has some ancient sigrok
<rqou>
wtf
<rqou>
i'm really on the verge of dropping debian
<whitequark>
technically, i think they have the latest release of sigrok
<whitequark>
so it's not the fault of debian
<whitequark>
but of sigrok for writing shitty code
<mithro>
daveshah: btw I would love you to work on micropython for FPGAs as part of FuPy - https://github.com/fupy - fupy.github.io
<rqou>
i just hate dealing with angry germans, broken gui stuff that i don't care about, and not-even-unstable-enough electronics/eda/cad/etc. packages
<balrog>
the version in sid is relatively current
<rqou>
hmm, i should check if they unfucked okteta yet
<Ultrasauce>
oh just to backtrack for a minute there rqou...if you're on debian that probably explains your past problems with gstreamer
<Ultrasauce>
downstream package fuckery abound
<rqou>
i used to be on ubuntu
<Ultrasauce>
worse for that
<rqou>
not sure if that helps or not
<whitequark>
i've yet to find a distro less shitty than debian tbh
<rqou>
arch? the weeb distro
<whitequark>
lol arse linux
<whitequark>
rolling releases are unusable
<whitequark>
really, anything that i can't safely update in cron is unusable
<rqou>
huh
<rqou>
i purposely switched to a rolling release because i started to find ubuntu unusable
<rqou>
every time i upgraded everything stopped working
<whitequark>
it's mostly for servers
<rqou>
and upgrades would always launch in the middle of the semester during crunch time
<whitequark>
the last time debian broke something on my server was like in 2008
<rqou>
yes, i run ubuntu on my servers
<whitequark>
which was the /dev/sda vs /dev/hda
<rqou>
and the last upgrade was a fucking disaster
<rqou>
i went from the old old LTS (14.04) to 16.04
<whitequark>
oh
<rqou>
and (initially) you couldn't even log in anymore
<whitequark>
you mean -dist-upgrades
<whitequark>
that's a different beast entirely
<rqou>
yeah
<rqou>
normal upgrades on ubuntu work great
<rqou>
the only problem is that their packages become stale AF
<whitequark>
why were you doing a dist upgrade in the middle of a semester during crunch time?
<whitequark>
this is stupid
<rqou>
i wasn't
<rqou>
i would defer it
<rqou>
i basically jumped ship after i had deferred an upgrade long enough that i couldn't install packages anymore
<rqou>
because the old stuff had been purged from the archives
<rqou>
but overall i'm not too happy with ubuntu but don't really see a better choice for servers
<rqou>
i wonder how much the upgrade to 18.04 will make me angry
<whitequark>
awygle: idle thinking
<whitequark>
on "next-glasgow" we might want to use SPDIF physical layer for sync port
<whitequark>
not sure how much does it actually matter
<rqou>
wtf
<rqou>
why?
<whitequark>
rqou: so that you don't have a common ground
<rqou>
aah
<rqou>
neat idea
<rqou>
the only problem is that i've never actually seen an spdif cable irl
<rqou>
make it both? :P
<whitequark>
"wtf why" → "neat idea"
<whitequark>
toslink cables are extremely cheap
<rqou>
"why not a SFP" <-- azonenberg :P
<whitequark>
no, that was azonenberg's idea, actually
<whitequark>
well it's derived
<rqou>
ooh yeah now i remember
<rqou>
wait...
<rqou>
i thought that was _my_ idea
<rqou>
for doing instrumentation for azonenberg's smart PSU/PDU
<whitequark>
oh, dunno, maybe
<whitequark>
i don't remember
<awygle>
I don't think isolated ground is a big deal for sync. More for duts imo
<awygle>
Toslink is very cheap though
<whitequark>
also it looks like you'll need a dual toslink cable or something
<whitequark>
since there are no combined transceivers
* awygle
briefly longs for a really good sound system
<awygle>
Isn't that a vendor thing? Aka not standard?
<awygle>
God I love marketing copy on high end av cables
<awygle>
I'm gonna start a tumblr
<whitequark>
seems so
<rqou>
whitequark: it's how they squeeze toslink into macbooks
<rqou>
which most people don't even know about until they run linux and encounter the bug where linux doesn't seem to be able to turn the toslink LED off
<rqou>
because apparently HDA Audio is a massive piece of junk
<whitequark>
lol
<whitequark>
does differential manchester encoding count as 1b/2b?
<rqou>
i guess?
<gruetzkopf>
hmm, spdif switching always worked for meTM
<whitequark>
gruetzkopf: btw i bought some 600W stew cooker for vapor phase soldering
<whitequark>
comes with a nice wire thingy for putting a PCB on
<gruetzkopf>
neat
<whitequark>
how do I get the heat transfer liquid off the PCB once I'm done?
<gruetzkopf>
for me tilting the pcb is enough
<gruetzkopf>
maybe shake it once
<gruetzkopf>
it really doesn't stick well to PCBs
<whitequark>
nice
<whitequark>
does it dissolve flux?
<gruetzkopf>
not really
<gruetzkopf>
takes a while to noticable aquire flux
<whitequark>
oh
<whitequark>
I can always distill it
<whitequark>
just simple distillation
<awygle>
do please report on how this goes, it sounds super cool
<whitequark>
awygle: oh I can send you some Galden too
<mithro>
rqou: I'm avoiding work by going through my RSS feeds
<awygle>
azonenberg: in your oven, do you put the metal trays in?
<azonenberg>
awygle: Yes
<awygle>
K
rohitksingh has quit [Quit: Leaving.]
<awygle>
azonenberg: how long after melt?
Bike has quit [Quit: Lost terminal]
<azonenberg>
awygle: I wait until all visible solder have melted, then turn off heat after 30 sec above liquidus then open the door
<awygle>
Mk that's basically what I did (45 sec after first melt, which is close to the same)
<azonenberg>
Yeah 30-45 is about right
<awygle>
Next time I need to center more carefully front/back
<azonenberg>
you had thermal gradients?
<azonenberg>
my oven is super uniform, the convection fan helps a lot with this
<awygle>
Little bit, not *too* bad but definitely visible
<azonenberg>
As far as all solder melting very close to the same time
<awygle>
Yeah this one has a fan
<azonenberg>
There's usually only a few seconds in which some stuff is melted and others isn't
<awygle>
I'm not sure which elements were on though
<azonenberg>
the one exception is big SMA connectors
<lain>
azonenberg: did you insulate your oven?
<azonenberg>
Those take time
<azonenberg>
lain: no
<awygle>
I set it to toast, not cookies :-P
<azonenberg>
it's literally stock
<lain>
interesting
<azonenberg>
awygle: yeah i looked in the manual
<azonenberg>
cookies has all elements on + fan
<azonenberg>
(on this oven)
<azonenberg>
Which is what i wanted
<lain>
we got a big convection oven and bolted a controller on, but measurements indicate we got a large thermal gradient with the fan on
<awygle>
I have an independent fan control knob
<lain>
turning the fan off resulted in more even heat
<azonenberg>
lain: interesting
<lain>
reading up online, people said it's a really good idea to insulate and add reflective tape inside
<azonenberg>
i got much better results with it on, though i want to fine tune the oven down the road
<lain>
interesting
<azonenberg>
My plan is to insulate with aluminum foil plus maybe glass wool or something
<azonenberg>
Build my own controller, not for profiling per se but i want to be able to push the temp 10-15C hotter to hit the optimal process temp for SAC305
<azonenberg>
Then put in a hose barb so I can feed N2 purge gas in
<whitequark>
awygle: btw why not vapor phase
<whitequark>
no need for N2 purge
<awygle>
There's no obvious indication of which elements are on in which mode, but if the bottom ones were on I feel like the top should be?
<azonenberg>
and then set up some kind of hood or exhaust system on top
<awygle>
There was a surprising amount of fumes though
<azonenberg>
To catch the fumes
<awygle>
Wonder if that's just because it's new
<azonenberg>
Yes, that is typical from all of the flux melting at once
<awygle>
whitequark: I had never looked into it until last week
<azonenberg>
Thats one of the reasons i wanted both inerting and exhaust in the new lab
<azonenberg>
inerting = less combustion of flux
<awygle>
azonenberg: yeah but even by that standard this seemed excessive
<azonenberg>
exhaust = outgassing is vented outside
<azonenberg>
how much? visible smoke comnig off the board?
<whitequark>
erm
<awygle>
is smol board
<whitequark>
azonenberg: why not vapor phase
<whitequark>
not awygle
<azonenberg>
whitequark: expensive, hard to obtain
<awygle>
Visible out of the top of the oven
<azonenberg>
in the US
<awygle>
Not near the board
<whitequark>
azonenberg: hard to obtain?
<azonenberg>
whitequark: in small volume, in the us? yes
<whitequark>
not at all
<whitequark>
sec
<azonenberg>
Biggest issue is, i have a convection reflow process dialed in
<azonenberg>
it works and i'm confident, i just want to tweak it a little
<awygle>
myyy apartment smells bad now
<azonenberg>
And while vapor phase seems cool, i don't see the benefits
<azonenberg>
for my use case
<awygle>
gotta work out a better system for exhaust
<whitequark>
oh hm, the supplier i was thinking of doesn't sell the HTF that works for lead-free
<whitequark>
so you might be right on that art
<awygle>
I overpasted this a bit but it's pretty solid. Just have to zap a couple shorts. Not bad for first board in 2+ years
<awygle>
(first board I assembled myself)
rohitksingh has joined ##openfpga
<azonenberg>
awygle: glasgow or something else?
<awygle>
else
<awygle>
This is my LNB board
<awygle>
Tomorrow is smolfpga
<awygle>
I'll post pictures in a bit, taking cat for a walk
<openfpga-github>
Glasgow/master ac2dad6 whitequark: Add Glasgow platform, programmer and target.
<openfpga-github>
Glasgow/master 0f351cd whitequark: Correctly use TSTriple in I2C gateware.
user10032 has joined ##openfpga
user10032 has quit [Remote host closed the connection]
user10032 has joined ##openfpga
* whitequark
ponders
<whitequark>
awygle: do you have an opinion on the license for Glasgow
<whitequark>
I'm tempted to go with my usual 0-clause BSD
<whitequark>
rqou: wtf why aren't you sleeping
<rqou>
good question
* rqou
zzzzzzzzz
<Ultrasauce>
sleep is for the weak
<Ultrasauce>
you're gonna do a run of those right? maybe I should procure one. been fantasizing about the sort of socratic dialogues with hardware it'll enable
<Ultrasauce>
a process currently limited by not wanting to bother writing firmware
<whitequark>
Ultrasauce: Glasgow? yeah
<whitequark>
socratic dialogue lol
<whitequark>
but I'm *definitely* going to first prove that I can pass timings now before making a production run
<Ultrasauce>
yeah theres certainly no point if the fpga is too wimpy
<whitequark>
I have several features in the design that make it possible to work with even real slow gateware
<awygle>
I would not use CC0 but I don't care very much and I consider this your board
rohitksingh has quit [Quit: Leaving.]
SpaceCoaster has joined ##openfpga
<whitequark>
awygle: what would you use?
<awygle>
whitequark: in the absence of external forces, probably OHL or CC-BY-SA
<whitequark>
awygle: and software?
<awygle>
GPL3 for the firmware, gateware same as hardware (but I haven't really dug into appropriate licensing for gateware)
* whitequark
looks at OHL
<awygle>
That's my biases though. CC0 is the best way to minimize the encumbrance which is my understanding of your preference
<whitequark>
"The Licensee [...] he [...] he [...]" ok fuck that noise
* awygle
winces
<awygle>
Fuck seriously
<awygle>
Get it together CERN
<whitequark>
I would modify it but "Anyone is welcome to use the CERN OHL, in unmodified form only, for the distribution of his own Open Hardware designs."
<whitequark>
>his
<whitequark>
lol nope
<awygle>
Yeah fuck that. I never noticed that (thanks privilege).
<whitequark>
awygle: I don't think I can license gateware same as hardware
<whitequark>
it's Python code
<awygle>
ahhh yeah
<awygle>
didn't think about that
<awygle>
like i said i haven't really investigated gateware much
<awygle>
i'm perfectly happy to go CC0 for the whole thing
<whitequark>
I mean, technically, that's Free, right
<awygle>
lmao
<awygle>
that's amazing
<Bike>
if the document is causeless sense data why would i follow its orders and denounce it as such
<Bike>
this will never hold up in court.
<whitequark>
hahaha
<whitequark>
but why would I care about it holding up in court?
<Bike>
good point
<awygle>
this is the first one of these "joke" licences i've actually respected
<whitequark>
it's srs bzns
<Bike>
given that my memories could be falsified as well, i think it's reasonable to say it's the only bzns.
<whitequark>
indeed
<felix_>
re cern ohl: wat?! o_O i wonder if or when they'll change that; i mean using a singular they/their works quite well in english. in german you can only either use plural or have some construct that makes your eyes hurt. will probably pick another license for my current open source hardware project; thought that i've solved the which license to pick question; didn't really read the license text though
<felix_>
tbh...
<cpresser>
i just looked up the cern ohl thing. it seems they changed it in version 1.2
<cpresser>
"... for the distribution of their own Open Hardware designs."
<openfpga-github>
Glasgow/master 81ea988 whitequark: Add the rest of FX2 pinout and LED control code.
<rqou>
0BSD _and_ Apache-2.0?
<whitequark>
yes
<rqou>
why?
<whitequark>
apparently some people care about patent clauses
<awygle>
Yeah Apache 2 is definitely the most biz friendly license
<whitequark>
I would obviously never take out a patent on anything in it but it also costs me nothing to add Apache-2.0 and it follows the spirit behind using 0BSD
<awygle>
whitequark: thanks for, uh... Giving a shit what license I prefer I guess? Lol
<awygle>
Unnecessary but appreciated
<whitequark>
well, I don't consider myself automatically right, it was worth investigating
SaLoMoN has quit [Ping timeout: 265 seconds]
<daveshah>
rqou: sorry to pester, but are you still planning to do those ice40 die photos?
<rqou>
yes
<rqou>
i _might_ get to them this week
<rqou>
but if not you'll probably have to wait for two weeks
<rqou>
upcoming week is dead week (no classes)
<rqou>
but week after that is final exams
<whitequark>
awygle: actually I don't care about businesses
<whitequark>
if they're scared of patents that's their problem, not mine
<whitequark>
I do care about academia though, and I was told Apache also makes things easier there "somehow"
<balrog>
> I was told Apache also makes things easier there "somehow"
<balrog>
offices of university counsel tend to be irksome, sometimes
<balrog>
afraid of any liability
<whitequark>
yeah that's a reason good enough
<daveshah>
rqou: sure, no worries
<daveshah>
just want to see them at some point :)
<awygle>
That makes sense to me yeah
<mithro>
whitequark: Normally it's the patent stuff which makes Apache 2.0 considered better
<whitequark>
yep
<whitequark>
that's why
* whitequark
stares at code
<whitequark>
so I'm using the same code paths to access EEPROMs and FPGA registers because the Cypress chip doesn't have a lot of code RAM
<whitequark>
I feel gross
<mithro>
whitequark: Yeah - the FX2 frequently makes you feel gross :-(
<jn__>
8051 :|
<awygle>
I constantly mock my father for the extremely dicklike shape of the rocket he works on, glad to see Twitter agrees
<rqou>
your father works for spacex?
<awygle>
Blue Origin
<awygle>
Google New Shepard if you haven't seen it
<mithro>
hey awygle - Any chance you'll get time to review those schematics today?
<awygle>
mithro: this evening yeah
<mithro>
awygle: Hust to be clear, I'm after review to make sure the PCB can be routed. I'm definitely expect there to be issues like missing junction dots, unconnected nets, etc.
<mithro>
awygle: I also haven't done a DRC check or anything myself
<awygle>
mithro: so more a feasiblility check than a detailed design review?
<azonenberg>
mithro: have you done ERC?
<whitequark>
grumble grumble kicad ERC is next to useless
<mithro>
awygle / azonenberg: the schematic was done by a person who is not a schematic designer
rohitksingh has quit [Quit: Leaving.]
<mithro>
So it should mostly "look right" but probably has issues like I mentioned above
* whitequark
is reminded of that Lattice FPGA devboard
<asy>
awygle: there is no shame to work on flying penis.
<whitequark>
blue origin should release company branded dildos as promo
<whitequark>
untapped market
<whitequark>
now the silicon valley libertarians could *literally* suck bezos' dick
<rqou>
lol
<whitequark>
after that elon musk couch thing i'm sure someone will do it, live, on youtube
<whitequark>
cheered on by hacker news
<whitequark>
there is not enough alcohol in the world to erase that mental image
<felix_>
mithro: hm, i wonder why do you want to make a new board with a spartan 6 on it?
<mithro>
felix_: in the repo?
<felix_>
yep
<felix_>
only looked at the description, not the schematics though
<felix_>
or are you planning to looking into continuing the work on the s6 fpgatools? ;)
<mithro>
felix_: Because the s6 is currently *super cheap* in China
<mithro>
felix_: We will see if I ever actually make the board, it's the lowest priority
pie__ has quit [Read error: Connection reset by peer]
pie__ has joined ##openfpga
<felix_>
ah, ok. do those cheap s6 still work properly? i'm a bit sceptical about salvaged chips
<balrog>
isn't the s6 pretty hated? :P
<mithro>
felix_: Seem too
<mithro>
I'm willing to risk it if I can get the boards in the $10 USD range...
<daveshah>
this using luke's bootloader?
<felix_>
oh, a board in the 10$ range with an fpga that isn't tiny would indeed be nice
<daveshah>
it would totally change perceptions about FPGAs
<daveshah>
obviously a board in the 10$ range with a big FPGA and open tools would be even better