<azonenberg>
but the emi fingers etc mena they wont pack as densely as i want
<azonenberg>
you'd need separate front panel cutouts for each
<awygle>
yup
<rqou>
azonenberg: so what about a bucket of copper SFPs?
<azonenberg>
rqou: out of the question
<gruetzkopf>
don't have my SFP to VGA adapter with me to check mine
<rqou>
gruetzkopf: wat
<rqou>
azonenberg: why?
<gruetzkopf>
i2c and a voltage reg off the 5V rail?
<rqou>
lol
<rqou>
but does it actually try to send VGA signals?
<gruetzkopf>
no
<rqou>
or just gives access to the i2c?
<gruetzkopf>
it's just a "wtf is this shit" adapter
<azonenberg>
awygle: oh, do we want to put some kind of ESD protection on the ethernet pairs too?
<azonenberg>
probably a good idea
<azonenberg>
gruetzkopf: sfp to vga? lol
<rqou>
azonenberg: wait does your favorite phy have TDR mode?
<awygle>
the phy has 8000 kV tolerance apparently
<azonenberg>
rqou: both the ksz9031 and the dp83867 have it
<azonenberg>
the TI part has better though, it can report up to 4 peaks vs the micrel's 1 and has more detailed info on amplitude etc
<azonenberg>
awygle: you think thats good enough as is then?
<rqou>
heh TI parts are usually surprisingly good
<awygle>
should be
<rqou>
azonenberg: MRVL parts :P
<gruetzkopf>
aaah no
<rqou>
azonenberg: i dare you to do a design around MRVL/QCOM/BRCM parts :P
* rqou
waits for azonenberg's head to asplode
<azonenberg>
rqou: even if i was able to get datasheets/silicon
<azonenberg>
i dont want to give them any design wins
<azonenberg>
them, plus Vitesse, are on my life-long blacklist
<azonenberg>
I will never do a design with one of their parts
<azonenberg>
Even for a commercial customer in high volume
<azonenberg>
s/even/especially/
<rqou>
what if somebody hires you to do consulting on one of those things? :P
<azonenberg>
I'd decline the gig
<rqou>
azonenberg: Huawei silicon :P
<azonenberg>
If the companies feel so strongly about not wanting my business
<azonenberg>
I'll give them what they want
<azonenberg>
And buy from a competitor :p
<rqou>
azonenberg: does this mean that all "secure element" vendors should be on your blacklist too? :P
<azonenberg>
i only blacklist a company once they've actually blown me off when i attempted to buy their product
<azonenberg>
so e.g. nvidia would probably end up on there if i tried
<azonenberg>
But i havent tried
<rqou>
but not infineon/nxp?
<rqou>
also, these are huge companies that make lots of parts
<azonenberg>
Yeah i'm aware, but MRVL/QCOM/BRCM/vitesse are like this for ALL of their products
<gruetzkopf>
yes
<gruetzkopf>
BRCM even for stuff they recently aquired
rohitksingh_work has quit [Ping timeout: 240 seconds]
<gruetzkopf>
ex-PLX datasheets are basically gone
<rqou>
oh wat
<rqou>
goddammit hock tan
<azonenberg>
For niche stuff like NXP secure elements i wouldnt blacklist the entire company since its not the bulk of their business
<azonenberg>
But i wouldnt buy any secure elements from them if i could avoid it
rohitksingh_work has joined ##openfpga
<gruetzkopf>
i've heard enough people loudly complaining about datasheet availability for them
<rqou>
nxp parts?
<gruetzkopf>
nxp "secure" stuff in particular
<rqou>
azonenberg: pwn a secure element when? :P
<gruetzkopf>
even as a potential 250kUnits/year customer
<rqou>
azonenberg: both karsten nohl and tarnovsky have done it before :P :P
<azonenberg>
i remember attempting to use an atmel atsha or atecc chip in a potential design
<azonenberg>
and having a hard time getting datasheets for those too
<rqou>
azonenberg we need more silicon pwns :P
<rqou>
azonenberg: we should call ourselves "Team V&" :P :P :P
<gruetzkopf>
MRVL is bad about this even for parts that are obsolete
<azonenberg>
gruetzkopf: the 88e1111 datasheet is *still* not public
<gruetzkopf>
for the processor in the steam link, there's conflicting information on CPU core count and implementer
<azonenberg>
how many years ago did it come out?
<azonenberg>
i mean sure its been leaked
<azonenberg>
But there's been no official release
<rqou>
it has
<rqou>
you can find it easily
<azonenberg>
yes i know, i may or may not have a copy :p
<azonenberg>
But even if i can find leaked documentation
<azonenberg>
morally, i don't want to give design wins to a company that actively fights my attempt to use their product
<rqou>
btw, what do you think the chances are that chip hackers get v&'d? :P
<rqou>
how many analysts are watching this channel?
<azonenberg>
I think if somebody like me is not at least on TLA radar, they need to get fired
<azonenberg>
That said, disappearing people is unlikely at least in the US
<azonenberg>
Especially if you're not targeting .mil chips etc
<azonenberg>
They're more likely to hire you and give you lots of money to keep your mouth shut and work in $AGENCY's lab :p
<gruetzkopf>
heh, i really need to record the current waveform going into this power supply
<gruetzkopf>
finest late-1950's grade three-phase to DC, 60V, 100A, technology
mumptai has joined ##openfpga
<rqou>
gruetzkopf: i bet the power factor will be amazing :P
<rqou>
does it involve mercury arc rectifiers?
<gruetzkopf>
naah
<gruetzkopf>
newfangled silicium rectifier
<gruetzkopf>
and exactly 2 transistors (not anywhere near the output path)
<rqou>
wow, modern :P
<gruetzkopf>
far too many power rails in this system (13)
<gruetzkopf>
but we do have power factor correction (and it always had)
<gruetzkopf>
theres' also a purely magnetic 3-phase->1-phase converter
<rqou>
huh
<rqou>
TIL that PFC had already been invented at that time
<sorear>
archaeo-PFC provides my favorite sense of the term "reactor
<gruetzkopf>
well, passive PFC in the form of a few huge capacitors
<rqou>
oh
<gruetzkopf>
though auto-switched to keep close to 1
<sorear>
100A for the train signal system??
<gruetzkopf>
it's overspecced a lot, so it can recharge it's batteries in finite time after power outage
eduardo__ has joined ##openfpga
<gruetzkopf>
i think we're currently at about 12A steady-state current on the DC side
<gruetzkopf>
(this is with ~10% of the 8000+ relays energised)
<rqou>
but where's the Internet of Targets? :P
<sorear>
ah, so 100A allows it to energize all of the relays at once
<sorear>
kinda surprised they didn't transistorize (*not* VLSI-ize) the whole thing in the 70s for MTBF reasons
<sorear>
don't relays break
<rqou>
gruetzkopf: how do I make all the signals green like in movies? :P
<sorear>
you put the IR flashy thing on the locomotive right?
<gruetzkopf>
there's still a lot of these in actual service, controlling some of the largest train stations in germany
eduardo_ has quit [Ping timeout: 264 seconds]
<gruetzkopf>
munich, cologne, dusseldorf are all this exact model, and many many smaller ones
<gruetzkopf>
transistorised things tend to not have a useful operational life of about 80 years
<rqou>
[22:51] (rqou) gruetzkopf: how do I make all the signals green like in movies? :P
<gruetzkopf>
it involves the right keys and a fair bit of system knowledge, and you'll get caught by the operator
<sorear>
discrete BJTs don't have an 80 year operational life?
<gruetzkopf>
there's axle counting systems with transistor-core logic, they have problems with transistors failing
<gruetzkopf>
try getting a provably-exact replacement for a 60s transistor
<rqou>
isn't there also a problem with overflow after 128/256 axles?
<rqou>
or is that the American system?
<gruetzkopf>
the transistor-core stuff overflows at 512 axles
<gruetzkopf>
(and will then keep the section marked as occupied)
<gruetzkopf>
the electromechanical counters used before are 256 or 384 steps/revolution steppers with a comparator between them, but those will also fail safe
<azonenberg>
awygle: I mostly agree with what he's saying, but there are times when I actually want to whole-hog replace an existing implementation
<azonenberg>
I prefer C++ plus restrictive coding style guidelines to a language that forces you into a square hole
<awygle>
Yeah I would want override as an escape hatch
<awygle>
With an explicit keyword
<rqou>
i personally prefer just not using methods at all
<rqou>
why not functions? :P
<azonenberg>
Basically, my general policy is that I know best in the end, but if I do something fishy i want to be called out on it
<azonenberg>
So thorough static analysis, heavy compiler warnings, etc
<awygle>
Galaxy brain: methods are functions its a useless distinction
<azonenberg>
But still have an option to say "yes i mean to do this"
<azonenberg>
Rather than fighting the language to do what i need
<azonenberg>
As you do with e.g. the Math class in java because you cant have functions that don't operate on state be global
<azonenberg>
so you make up a useless container class just so they can go somewhere
<rqou>
yeah, java-brand OO is gross
<azonenberg>
I want a language that gives me tools
<rqou>
but in general i just don't prefer to use inheritance
<azonenberg>
but lets me use them as i see fit
<azonenberg>
Sure, a hammer is harder to kill yourself with than a ramset gun
<azonenberg>
It's the best choice in 95% of cases
<azonenberg>
but sometimes you actually need to drive a nail an inch into concrete
<azonenberg>
Which is why I like "C+"
<rqou>
or Rust? :P
<azonenberg>
I havent had the time to learn rust yet
<azonenberg>
what i can say is, even if i did
<azonenberg>
it wouldnt replace C++ for me
<rqou>
because you're constantly messing with your house :P
<azonenberg>
i'd use it in certain niche applications
<azonenberg>
Lol
<azonenberg>
We got the old carpet off the stairs today
<azonenberg>
threw out another dumpster load of insulation and sheetrock debris
<azonenberg>
and did a lot of "detail demo" pulling nails and trim and other stuff
<rqou>
wtf
<azonenberg>
And swept and vacuumed the floors upstairs several time
<rqou>
when are you ever going to get to actual installing stuff?
<rqou>
not just ripping stuff out?
<azonenberg>
we literally spent a month removnig insulation
<rqou>
why?!
<azonenberg>
blow-in insulation is a nightmare
<rqou>
oh
<azonenberg>
there's no good way to remove it
<rqou>
ugh
<azonenberg>
you can shovel it, or vacuum it
<rqou>
i'm sorry
<azonenberg>
or some combination thereof
<azonenberg>
It's made worse when it's in tight corners
<rqou>
wait, is this urethane foam insulation or something else?
<azonenberg>
No
<azonenberg>
About ten inches of fiberglass "cotton ball" pieces on top of four inches of fire retardant sawdust
<rqou>
"fire retardant sawdust"?
<rqou>
that's possible?
<azonenberg>
they treated it with something apparently
<azonenberg>
with all the roof leaks it's too soggy to burn anyway :p
<azonenberg>
(which makes it doubly fun to remove)
<rqou>
have you at least gotten the roofers to show up yet?
<azonenberg>
anyway to get rid of it i'm in a climbing harness, resting my weight on two 2x4 pieces stretched across the ceiling joists
<azonenberg>
pressing uncomfortably in various parts
<azonenberg>
i cant move more than a few feet without changing my tie-off to another anchor point, since i'm only 8ft above floor level i cant use a very long rope
<azonenberg>
on my belly
<azonenberg>
trying to shove myself into a spot so low (on the edges of the roofline) that i have to turn my head sideways
<azonenberg>
because head + hardhat wont fit upright
<rqou>
have you tried hiring somebody instead? :P
<azonenberg>
holding the hose for a vacuum cleaner shoving it even further into the corner
<azonenberg>
suckign it out
<azonenberg>
after 5-10 minutes of this, the 50-gallon bucket is full
<azonenberg>
and i have to roll it down the ladder and wait for ally to shovel it out
<azonenberg>
When we started, we didnt think it was gonna take this long
<azonenberg>
and now we're 99% done
<azonenberg>
we just have one last tihgt spot over the kitchen ceiling
<azonenberg>
Let's just say it's a very good thing i am not claustrophobic
<rqou>
and then you have all the actual _installing_ to do
<azonenberg>
We've been wiring as we do demo
<azonenberg>
And the demo isnt the only thing, there's a lot of decontamination involved
<azonenberg>
everything is loaded up with tobacco residue
<azonenberg>
Today, we cleaned the floors in most of the second floor
<azonenberg>
This involves sweeping the bulk debris up first
<azonenberg>
that are too big to vacuum
<azonenberg>
Then going over the whole floor with a shopvac and hepa filter
<rqou>
have you considered just demo-ing the whole house? :P
<azonenberg>
Then making a second pass with a broom and "sweeping compound", which is basically sawdust or some other absorbent material soaked in some kind of oil
<azonenberg>
it binds to particulates and keeps them from becoming airborne
<azonenberg>
great for getting fine dust off a floor, especially an uneven one
<azonenberg>
Then we do a second hepa vac pass to remove traces of the compound (since it doesnt all stick to the broom when you sweep) and additional dust
<azonenberg>
We did the first 2 on the whole house, two bedrooms still need the last two passes
<rqou>
seriously wtf dude
<azonenberg>
Then we're going to wait until the first floor is a bit more finished
<rqou>
should have just demo-ed the whole house and started over :P
<azonenberg>
And then get a spray bottle and vinegar, which apparently does a good job of degrading tobacco tars etc
<azonenberg>
And scrub every ceiling joist
<azonenberg>
every stud
<azonenberg>
and every inch of floor
<azonenberg>
Possibly more than once
<azonenberg>
And honestly, if buying a plot of land and building was an option i probably would have
<azonenberg>
Knowing what i do now, i might - but we made the best decision with the knowledge we had at the time
<azonenberg>
At the tmie we believed that we had to be out by today
<azonenberg>
time*
<azonenberg>
We later found out the landlord's deployment was being extended, so we had more tmie
<azonenberg>
We basically knew we had to buy something asap because we were getting kicked out
<azonenberg>
Building was not an option in that time frame
<azonenberg>
If we knew we'd have until august, i might have saved up a bit more, bought some land, then built to suit
<azonenberg>
However, that brings me to the second issue which is that buildable land is scarce around here
<azonenberg>
and buying a house, knocking it down, then building another one is pricey
<rqou>
what about just living with monochroma? :P
<rqou>
not "real adult" enough for you? :P :P :P
<azonenberg>
That would entail putting all our stuff in storage since she doesnt have enough space for my lab etc
<azonenberg>
b) being a lot more extensive with packing, since everything would have to be boxed vs carying stuff over a bit at a time in the car then having a mover take only the big stuff
<azonenberg>
c) extended downtime
<azonenberg>
awygle: yeah i read that paper when it came out
<azonenberg>
Scary :p
<azonenberg>
(have i mentioned we need open source chip layout yet?)
<rqou>
magic/qflow?
<azonenberg>
no i mean
<azonenberg>
have an industry policy that your GDS is provided as part of the SDK
<azonenberg>
:P
<rqou>
lol not happening
<awygle>
Permitting is also a nightmare for building, or so people keep trying to convince me
<rqou>
even parallax never gave away their gds and afaik they fully control it
<rqou>
awygle: build illegally? :P :P :P
<awygle>
I still think I'm more likely to build than buy, but we'll see in a couple years
<awygle>
Gotta get un-poor :-P
<rqou>
shoulda bought bitcoin :P :P :P :P :P
<awygle>
(actually I'm already quite un-poor but houses are expensive)
<rqou>
get a roommate? :P
<azonenberg>
awygle: well fwiw
<azonenberg>
The guys who renovated this house before me
<awygle>
I almost certainly will, when/if I have a house
<azonenberg>
are the reasons building permits exist :p
<azonenberg>
i'm finding so much BS
<azonenberg>
and the scary part is, the things i'm able to call out as awful workmanship are mostly electrical in nature
<azonenberg>
Because i know the NEC decently well
<rqou>
awygle: PNW openfpga dorm? :P
<awygle>
azonenberg: yeah but apparently on the east side you get at least some of the bay area "this area is totally zoned for this" "but my petunias!" horseshit
<azonenberg>
How many more awful things are hiding where i simply dont know the relevant part of the building code? :p
<azonenberg>
awygle: lol
<azonenberg>
... oh
<azonenberg>
that
<azonenberg>
well bainbridge just put in a moratorium on all building permits for a while
<awygle>
Doubt it's a problem in kitsap
<azonenberg>
curious how that will shake out
<azonenberg>
("just" = a few months ago)
<rqou>
why do we not have an openfpga dorm yet?
<azonenberg>
rqou: lol
<awygle>
Because we all have lives and better things to spend money on
<awygle>
Y'all can stay with me anytime, just don't make me your housing plan :-P
<azonenberg>
rqou: well i wouldnt be interested anyway, seeing as i already have a roommate :P
<rqou>
we need to get you another roommate to guarantee no future crotchfruit timesinks :P :P :P
<azonenberg>
Who i happen to be married with, but that's just a minor detail
<azonenberg>
lol
* awygle
has lived in a place the size of azonenberg's current place that had 7 tenants
<rqou>
lol
<rqou>
we used to have 6 here
<rqou>
but it's been whittled down to only 4 now
<azonenberg>
awygle: My last place before i moved out here was a 2500 ft^2 or so apartment with five bedrooms
<azonenberg>
fairly small, maybe 100sft each
<azonenberg>
ft2*
<azonenberg>
(how did that come out?)
* awygle
luxuriates in his 900 Sq ft apartment
<azonenberg>
With five tenants
<azonenberg>
Then a massive common area / dining room / living room
<azonenberg>
Of course i turned most of that into my lab :p
<azonenberg>
We still had a place to sit and eat, and we still had a couch/tv area
<awygle>
My lease is up in a couple months... Sigh. Hope my building doesn't try to fuck me too hard. There are three new buildings opening up this summer.
<rqou>
lol you don't need a place to sit and eat, nor do you need a couch/tv area :P
<rqou>
just lab space everywhere :P
<azonenberg>
rqou: that was their demand, not mine
<azonenberg>
i made it clear to my roommates that if they had a legitimate plan to use any of that vacant space, i would let them have it
<azonenberg>
But as long as nobody else wanted the common space, it was mine
<awygle>
Arright enough of this. Goodnight all
* awygle
zzz
<azonenberg>
i kinda miss that place sometimes
<azonenberg>
$575/mo rent
<azonenberg>
including all utilities
<azonenberg>
walking distance to everywhere i needed to go
<azonenberg>
rqou: also re timesinks
<azonenberg>
What about crotchfruit lab assistants?
<rqou>
nope
<azonenberg>
an in-house smt assembly tech would be very handy
<azonenberg>
:p
<rqou>
only labcats for me :P
<azonenberg>
Especially if [s]he doesn't need to be paid because it's their daily chore
<rqou>
but you still have to deal with all of the other problems
<rqou>
but lab cats just need to be fed and entertained :P
<azonenberg>
Nerdlings don't need litter boxes that stink up the whole house
<azonenberg>
Sure, diapers stink but at least it's not for their whole life
<azonenberg>
And lab cats can't build PCBs
<azonenberg>
And they're not as cute as lab puppers
* azonenberg
hides
<rqou>
fine, when openfpga manages to fork off opencrispr we'll practice on your crotchfruit :P
<azonenberg>
lol
<azonenberg>
no genetic experiments tyvm
<azonenberg>
now, if the process was already well developed...
<rqou>
no catgirls for you? :P
<azonenberg>
ally would probably be thrilled to have a catgirl kid :p
<rqou>
opencatgirl :P
<rqou>
hmm, just wondering, are gender-selective abortions legal in the united states?
<rqou>
they're illegal in china/india
<azonenberg>
Assuming you found out before deciding to abort, i'm not sure there would be any way to prove in court that it was / was not a factor in your decision
<azonenberg>
But i havent looked for case law on the subject
<rqou>
afaik in china you just aren't allowed to test for the sex of the fetus
<azonenberg>
ah ok
<azonenberg>
in the us it's common to find out during routine diagnostics, although idk if that would be past the point you'd consider terminating it
<azonenberg>
probably varies state to state
<rqou>
yeah, sex-selective abortion is totally legal and has been defended by the courts in the US
<azonenberg>
Welp, TIL
<rqou>
don't forget that we have the GOP :P
<azonenberg>
how about species selective? are you allowed to check for cat ears?
<azonenberg>
:p
<rqou>
hmm apparently the law does vary by state in the US
<rqou>
just like the difficulty of getting an abortion varies
<azonenberg>
i.e. level of republican domination varies
<azonenberg>
:p
<rqou>
yes
<rqou>
there's a paper somewhere on law.cornell.edu that claims "However, the text of the laws and the statements made in support of the bans during legislative hearings make it clear that they are intended to place restrictions on abortion services generally. Moreover, the laws purport to solve a problem that may not exist at all in the United States. Rather than changing behavior or addressing a purported problem,
<rqou>
sex-selective abortion bans are likely to lead to the denial of health care services to Asian American women."
<rqou>
which is typical republican bullshit as usual
<azonenberg>
How's that for a 20 minute schematic? lol
<rqou>
your schematics always look a bit weird to me
<rqou>
e.g. why aren't there GND symbols?
<azonenberg>
I just use net names for everything
<azonenberg>
GND is ambiguous
<azonenberg>
do you mean chassis ground? analog ground? digital ground?
<rqou>
and why is there a big empty J1I?
<azonenberg>
Because it was easier to make all instances of the 8-port rj45 the same size
<azonenberg>
and i had to hook the shield up somewhere
<azonenberg>
So i made a 9th subcomponent with just a ground pin and didnt bother to resize it
<rqou>
ooh this is a 2x4 thing
<rqou>
right
<azonenberg>
Yes
<rqou>
i basically never use multiple units
<azonenberg>
i'm using this board as a footprint verification for the line card too
<azonenberg>
i had planned to do that anyawy
<azonenberg>
I'm just hooking up test points too
<rqou>
i hate multiple units
<azonenberg>
try doing a big fpga
<rqou>
at least it doesn't have implicit vdd/gnd
<azonenberg>
and you'll hate not having them
<azonenberg>
i consider implicit vdd/ground a terrible idea and nevr use them
<azonenberg>
I am very explicit, for big fpgas sometimes i have an entire subcomponent for just vcc and another for ground
<azonenberg>
some of the kicad lib symbols stack pins
<azonenberg>
i NEVER do that
<rqou>
some of the 74xx parts still use implicit vdd/ground
<rqou>
i do use pin stacking
<azonenberg>
Which is why i use my own 74xx symbols
<azonenberg>
They have to keep the old ones to avoid breaking old schematics
<rqou>
usually on "potato" parts
<azonenberg>
But they're deprecated for new ones
<rqou>
avoid breaking old schematics?
<rqou>
what?
<rqou>
schematics aren't self-contained?
<whitequark>
no they aren't
<rqou>
wtf
<azonenberg>
They link to library symbols
<azonenberg>
PCBs include components though
<whitequark>
they do have the complete symbol inside them
<rqou>
but that has so much footgun potential
<azonenberg>
It does, and that's being fixed in the next version as part of the schematic revamp
<azonenberg>
whitequark: i think that may be a recent change
<rqou>
eeschema-new when? :P
<whitequark>
when you open a schematic and the symbol is updated, kicad asks you if you want to update the symbol or to extract it into a "rescue" library and link to that
<azonenberg>
earlier on, they didnt even have a cache
<whitequark>
oh yeah
<rqou>
whitequark: oh yeah, i got that message
<azonenberg>
the sch ONLY had the symbol name
<azonenberg>
i have old schematics i cant open anymore
<azonenberg>
they just show up as a ??
<rqou>
i've been ignoring it for now because i don't actually need to update my old schematics at this point
<rqou>
just quickly glance at one
<whitequark>
I make pdfs for that
<rqou>
also, seriously kicad schematic symbols are garbage
<rqou>
e.g. they've given themselves a fun backcompat issue
<rqou>
and they look so excessively "euro"
<whitequark>
that's a good thing
<whitequark>
fuck the US
<azonenberg>
i've grown to like the boxy resistors
<azonenberg>
vs the zigzag style
<whitequark>
rqou with his classic "it doesn't fit my taste therefore it's garbage" routine
<whitequark>
do you ever look at yourself from a distance?
<rqou>
you're basically doing the same thing
<whitequark>
i'm using the kicad libraries and contributing to them ...
<whitequark>
so no i don't think so
<rqou>
meanwhile i just trust azonenberg to get it right
<rqou>
have you assembled glasgow yet?
<whitequark>
don't have any solder paste
<whitequark>
i hate rs components, they always promise "1-day delivery" but in reality it's "3 working days" and then they're usually 1-2 days late on top of that
<rqou>
anyways, i just don't agree with kicad's "our way or go away" library conventions
<whitequark>
having a ton of inconsistent symbols is worse
<rqou>
i just don't get why you care so much about "consistency"
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
<rqou>
if it works and is understandable, ¯\_(ツ)_/¯
<whitequark>
"it works" is the lowest possible bar
<whitequark>
the "fuck it, ship it" school of engineering
rohitksingh_work has quit [Ping timeout: 248 seconds]
<jn__>
pie__: printing monospace text with fg/bg colors isn't hard
<pie__>
kinda wish i could find the game but i have absolutely no idea how so eh.
<jn__>
DEC VT emulation is hard, but you don't need that in a closed system such as a game
<pie__>
yeah
<pie__>
though with how slow i am (and not that i actually have any material) it might be possible to wait for alacritty to mature a bit, then id probably have a bundleable tty
<pie__>
assuming they get the "alacritty as a library" stuff working
m_t has joined ##openfpga
<pie__>
basically i want to be able to sanely draw "text", not just pushing lines at the bottom
<pie__>
if thats a sane thing to want to do
<pie__>
.me just got a severe urge to put some GITS music on
<jn__>
pie__: hmm, GITS
Bike is now known as Bicyclidine
<pie__>
gotta get that CYBER feeling
<gruetzkopf>
DEC VT is easy
<gruetzkopf>
go to your storage facility and grab a new-in-box DEC VT :P
<jn__>
:D
<gruetzkopf>
or how about a tandberg data TDV-2200 ECMA48-germany one
Bike has joined ##openfpga
<pie__>
is vcpkg Yet Another Package Manager but from microsoft this timee?
SpaceCoaster has quit [Ping timeout: 240 seconds]
SpaceCoaster has joined ##openfpga
ondrej2 has quit [Quit: Leaving]
m_t has quit [Quit: Leaving]
<jn__>
TIL: mIRC's scripting language is documented on wikichip
<rqou>
felix_ is getting somebody to work on those
<balrog>
ah
<rqou>
oh wtf
<rqou>
esp-open-sdk also has a broken AF build system
<cr1901_modern>
The entire C toolchain is... bad. I just use micropython, which well... actually works fairly well ._.
<rqou>
e.g. "make -j20" doesn't work
<rqou>
and it somehow managed to poke my PATH and find the totally wrong version of gcc
<rqou>
i don't get it
<rqou>
musl-cross-make manages to "just work"
<gruetzkopf>
IOA still always matches to international oversight authority for me. yay stargate.
* awygle
watched the movie, SG-1, and Atlantis, and does not recognize this TLA
<gruetzkopf>
it's the organisation that employed woolsey during sg1
<azonenberg>
awygle: we have "advisory services" which is kinda like that
<azonenberg>
and i've done onsite consults with clients for things like architecture reviews
<rqou>
balrog: is the esp32 bootrom hidden?
<rqou>
oh wtf esp32 also has bt classic in addition to BLE?!
<rqou>
how did they cram so much stuff into this chip?
m_w has joined ##openfpga
<azonenberg>
awygle: btw did you see the crosstalk characterization board i linked last night?
<azonenberg>
any thoughts on it?
<azonenberg>
The latest version has 950 ohm resistors on the test points after the differential terminators, so you can run 50 ohm coax to your scope and have a 20:1 integrated Z0 probe
<azonenberg>
(incidentally, if you don't hook up the aggressor then this can be used as a test platform for general ethernet signal quality measurements)
<azonenberg>
i kinda wanted one of those anyway
[X-Scale] has joined ##openfpga
X-Scale has quit [Ping timeout: 256 seconds]
[X-Scale] is now known as X-Scale
<awygle>
azonenberg: i saw the link but didn't click it, i'll get there
<whitequark>
bfd doesn't have a crash issue, it has a perf issue
<whitequark>
and even gnu people don't want to touch ld.bfd
<whitequark>
that's why gold exists
<rqou>
wait, we're only talking 4 seconds total?
<rqou>
meh, whatever
<whitequark>
that's a test case...
<whitequark>
not llvm
<rqou>
oh
<rqou>
um...
<kc8apf>
multi-hour link times are pretty common on big code bases
<rqou>
yeah I know
<whitequark>
that's more about lto
<awygle>
linking da worst
<awygle>
OOMing my machine alla da time
<whitequark>
llvm with split-dwarf and gold links every binary in a few seconds, 3-5 maybe
<whitequark>
and shared library builds of course
<kc8apf>
Photoshop used to run out of address space with 32-bit linkers
<whitequark>
firefox runs out of AS with 32-bit linkers for many years
<rqou>
Firefox too afaik
<whitequark>
maybe like a decade at this point
<whitequark>
and those were just symbol names
<whitequark>
wait no
<whitequark>
the symbol names thing was webkit
<rqou>
solution: stop using 32-bit linkers
<whitequark>
C++ is one hell of a drug
<rqou>
seriously, there's no reason to build 32-bit x86 for any "serious" software anymore
<kc8apf>
rqou: also postpone touching debugging info
<rqou>
even my old old machine from 2006 supported x86_64
Bike has quit [Ping timeout: 260 seconds]
<rqou>
(but yes, it shipped with a 32-bit winxp)
<whitequark>
microsoft still uses the 32-bit toolset i think
<kc8apf>
Ubuntu Bionic doesn't have 32-bit x86 installers
<kc8apf>
discovered that when I couldn't get icecube2 to work
<rqou>
i thought Microsoft also had a toolset that ran on x86_64 but targetted x86?
<whitequark>
yes, they do
<whitequark>
both their toolsets target x86 and x86_64
<whitequark>
kc8apf: apt-get install multiarch-support or something
<rqou>
oh they actually have all permutations now?
<kc8apf>
yeah, I got annoyed at that
<kc8apf>
too many libs to track down
<kc8apf>
it'd be better if Lattice shipped a .deb
<rqou>
some ancient versions didn't have every single permutation (but also had itanic support)
<rqou>
(unpopular opinion) it'd be better if they shipped a static binary
<awygle>
ubuntu dropped ia32-libs
<awygle>
i had to apt-get install them all manually for icecube2
<rqou>
only the meta-package though, right?
<awygle>
and then i couldn't get it to take my license file (which i think is just because i believed the filthy liars when they said i could browse to it)
<awygle>
i was pleased that alien turned the rpm into a deb without issue though
<rqou>
lol alien
<whitequark>
awygle: wtf ubuntu
<rqou>
(unpopular opinion) tarballs
<whitequark>
i'm never using it again then
<awygle>
tarballs are (mostly) fine
<whitequark>
ia32-libs are a requirement
<balrog>
rqou: pacman packages are essentially tarballs
<rqou>
I'm pretty sure they still have the actual packages
<awygle>
whitequark: yeah i think it was in 17.04 or something like that
<kc8apf>
my goal is to handle all the known architectures
<whitequark>
eech, nom
<kc8apf>
I get away from nom quickly
<balrog>
more rust code :P
<awygle>
meanwhile my goal is to take that support and work on architecture-agnostic tools like packing, p&r, etc
* balrog
looks at azonenberg
<awygle>
(... yknow, someday)
<whitequark>
awygle: niiice
<whitequark>
awygle: btw
<whitequark>
do you think you can fix the rest of footprints?
<whitequark>
or should I
<rqou>
meanwhile xc2bit just has magic proc macros :P
<awygle>
whitequark: yeah i plan to do that tonight, this weekend just got away from me a bit
<whitequark>
alright, thanks!
<awygle>
thanks for keeping me honest lol
<awygle>
whitequark: i'd love to hear your thoughts on the general gaffe concept. kc8apf and i have talked about it as "LLVM for FPGAs", but you're, like, a real LLVM dev :p
<whitequark>
uhhhhm
<whitequark>
it's a comparison that's too high-level to give any technical insight
<awygle>
sure, that wasn't really what i meant
<whitequark>
the only part of llvm that might translate to gaffe is the pass architecture
<whitequark>
so you might want a common IR then
<whitequark>
but... that's what yosys does already
<whitequark>
I guess gaffe is more like LLVM's MC layer?
<whitequark>
we still get 240 Mbps with the 30 MHz interface, so it's not *too* bad...
<whitequark>
but I'm worried about the other side
<awygle>
yeah but it is sad
<whitequark>
there's no way we can make an MPSSE run at 50 MHz
<awygle>
hmm yes
<whitequark>
well
<whitequark>
there's DDR
<awygle>
you can go the 2232D route
<whitequark>
oh?
<awygle>
that would give you 12 MHz instead of 60 MHz
<whitequark>
no, that's not it
<whitequark>
our frontend allows 50 MHz
<awygle>
oh, yes
<whitequark>
so we must run at least at 50 MHz
<awygle>
that would have to be DDR
<whitequark>
actually, fully utilizing our frontend would be more like 50 MHz DDR.
<whitequark>
well.
<whitequark>
it's settled then, we have to go for HX8k
<awygle>
it does seem that way
<rqou>
lolol
<whitequark>
can you make a symbol? :)
<awygle>
it would make sense to me to build out the whole pipeline, at reduced speeds, on the current revision and make sure we can make timing on the HX
<whitequark>
sure, I agree
<awygle>
if that isn't going to involve too much hoop jumping
<whitequark>
I'll make a very small run of UP5K devices
<awygle>
whitequark: would you be interested in continuing to do the firmware/gateware and having me do the bulk of the rev B PCB work?
<whitequark>
awygle: yeah that would really help
<awygle>
whitequark: okay, will do
<awygle>
trace and space is gonna be... interesting
<whitequark>
really? the package has 190 or something balls and we need like 48
<awygle>
yeah, there are 60 exterior balls in fact
<awygle>
power is the thing
<awygle>
i haven't checked yet but i bet the power balls are centrally located