<rqou>
you know the software you're trying to pwn is some fancy enterprise stuff when it's written in java and all google searches are SEO'd to hell with people trying to sell you training
<pie_>
isnt that most of all google searches these days anyway
<pie_>
the web used to be cool
<rqou>
some google searches do still work
<rqou>
but yeah, lots of things are worse
<rqou>
e.g. if you have a Windows error, god help you nowadays
<rqou>
"what do you mean I'm not supposed to fix errors by downloading a random dll from the internet?" /s
<jn__>
choose wisely between these five download buttons
genii has quit [Remote host closed the connection]
<qu1j0t3>
jn__: Ha!
<qu1j0t3>
yeah it's a cesspit. and I know exactly the dark corners of Java land rqou is talking about. anything to do with OData for examle
s_frit has quit [Remote host closed the connection]
s_frit has joined ##openfpga
ayjay_t has quit [Read error: Connection reset by peer]
ayjay_t has joined ##openfpga
<rqou>
why is enterprise software always so inner-platform-y?
<qu1j0t3>
ha.
<awygle>
I tried to google something about COM today and Google straight refused to use that search term. I put it in quotes and everything and just nothing relevant
<qu1j0t3>
then they name their language `Go'.
<rqou>
have fun with all the IUnknown :P
<qu1j0t3>
rqou: I expect Conway's Law has a lot to do with your inner-platform observation.
unixb0y has quit [Ping timeout: 252 seconds]
unixb0y has joined ##openfpga
florolf has quit [Ping timeout: 252 seconds]
florolf has joined ##openfpga
rohitksingh has joined ##openfpga
ayjay_t has quit [Read error: Connection reset by peer]
ayjay_t has joined ##openfpga
rohitksingh has quit [Quit: Leaving.]
rohitksingh has joined ##openfpga
rohitksingh has quit [Quit: Leaving.]
mumptai_ has joined ##openfpga
mumptai has quit [Ping timeout: 272 seconds]
zkms has quit [Ping timeout: 264 seconds]
zkms has joined ##openfpga
Bike has quit [Quit: Lost terminal]
rohitksingh_work has joined ##openfpga
ayjay_t has quit [Read error: Connection reset by peer]
ayjay_t has joined ##openfpga
<awygle>
hm. okay so i had a dry pad on the input capacitor for a regulator, and now it's regulating _way_ high for some reason (like a volt high on both output rails). does that make sense, that that could have damaged the regulator in this way, or do i have another problem?
<azonenberg_work>
awygle: is it oscillating maybe?
<azonenberg_work>
feedback loop going unstable?
<azonenberg_work>
and are you probing the output open circuit or under load
<awygle>
azonenberg_work: doesn't really look like it...
<awygle>
under load
<azonenberg_work>
thats really weird
<awygle>
also it's a fixed output LDO so i can't even have effed up the feedback resistors
<azonenberg_work>
What voltage are you getting vs wanting?
<awygle>
it should be 3.3V and 1.2V outputs
<awygle>
i'm getting like, 4.4V and 2.3V or something close, let me check
<awygle>
4.6V and 2V
s_frit has quit [Remote host closed the connection]
<zkms>
is the input voltage OK
s_frit has joined ##openfpga
<awygle>
yes, right on 5V just as expected
<awygle>
and the ground pin seems connected and so does the thermal pad (although it's hard to really see)
<azonenberg_work>
huh
<zkms>
is the input current at 5V nominal or weird?
<azonenberg_work>
Did you have zero capacitance on the input or what?
<awygle>
i thought it might be an incorrect part number but that doesn't seem to make sense
<azonenberg_work>
were there other caps on the 5V rail further away?
<awygle>
azonenberg_work: probably, the pad was totally dry. there _aren't_ any other caps on 5V proper actually although there's one on VBUS (so before the protection IC)
<awygle>
zkms: i don't have a good way to measure that at the moment
<azonenberg_work>
Hmmm, i wonder if you had issues with ringing or something on the usb port
<azonenberg_work>
inductance of the cable i mean
<awygle>
could be, i suppose
<azonenberg_work>
inrush current then keeps going and flies way above 5V?
<zkms>
yeah can you stick a scope probe on the 5V line
<azonenberg_work>
the capacitance would normally damp that out
<zkms>
or on the ldo outputs
<awygle>
well i put the cap back on, so this is persisting past that point. has to be damage to the LDO or a problem downstream that's making it oscillate _very_ fast
<awygle>
i'll get my scope out
<awygle>
hm. the 1.2V output _might_ be oscillating but it's oscillating within like 200 mV of 2.3V. the 3.3V output looks rock steady at 4.6V
<awygle>
i guess maybe i should try replacing the LDO...
<awygle>
same with input, looks fine
<azonenberg_work>
Got a spare? if so, thats a quick check
<azonenberg_work>
out of curiosity what was downstream of that LDO? it probably fried too :(
<awygle>
i do, and uh... _literally everything_ :/
<azonenberg_work>
I meant on what board
<azonenberg_work>
a glasgow?
<awygle>
yeah
<zkms>
can you hook up the LDO when not on the board to just 5V and capacitors and a bit of load?
<awygle>
maybe. it's an 8-pin MSOP with a thermal pad... i can hook it up on a blank board though once i remove it
<pie_>
do you have another regulator?
<pie_>
oh right after everyone else alreads said that, oh well :P
<pie_>
nooo :c <awygle> i do, and uh... _literally everything_ :/
<azonenberg_work>
awygle: good news is, momentary overvoltage may not hav ekilled things
<azonenberg_work>
you can actually get away with a fair amount for a short time
<azonenberg_work>
It will likely fail prematurely but a small overvoltage is usually not instantly fatal
<azonenberg_work>
So if your board dies of electromigation in 3 years instead of 10 blame this
<pie_>
didnt he have consistent overvoltage though
<pie_>
oh well idk what glasgow expects though, is 30% overvoltage a lot? :P
<azonenberg_work>
pie_: momentary meaning a couple of minutes
<azonenberg_work>
vs hours or days
<pie_>
oh
<pie_>
i thought magic smoke is generally pretty fast
<azonenberg_work>
Depends on how high and where its applied
<pie_>
makes sense
<azonenberg_work>
inputs > Vdd are problematic if you forward bias the protection diode
<azonenberg_work>
that will blow out the diode pretty fast
<azonenberg_work>
But if Vdd is too high as well, the biggest thing is probably gate oxide damage
<azonenberg_work>
if you're only a little too high you will get degradation but it probably wont arc over
<azonenberg_work>
(which is an insta-kill, of that fet at least)
<azonenberg_work>
then higher current everywhere so more electromigration
<azonenberg_work>
but thats also a slow, cumulative death
<azonenberg_work>
Overvolting by just a little bit will exponentially reduce the lifetime for sure
<azonenberg_work>
a 3.3V part at 4.6V might well survive a couple minutes at least
<rqou>
i assume it's not as bad as incandescent bulbs? :P
<azonenberg_work>
1.2V at 2.3V is more likely to be instantly lethal
<azonenberg_work>
in terms of percentage over the design Vdd
<pie_>
makes sense
<rqou>
iirc incandescent bulb lifetime shortening was ∝ V^4 or something
<azonenberg_work>
but it all depends on how much safety margin is built into the design
<pie_>
(well i suppose the converse would also "make sense" if someone said it to me ¯\_(XD)_/¯ )
<pie_>
right
<azonenberg_work>
for example, its totally possible that the "1.2V" process will survive for a week at 2.2V
<rqou>
fwiw i've overvolted a s3e to 5V on a vddio bank for a few seconds and it survived ¯\_(ツ)_/¯
<azonenberg_work>
but at 1.2 it will last 20 years
<sensille>
i thought it was mainly a problem of heat dissipation
<azonenberg_work>
sensille: Higher voltage leads to higher heat, yes, but heat doesnt kill chips directly
<azonenberg_work>
Higher voltage also leads to higher current
<azonenberg_work>
Electromigration in wires is acceleterated, exponentially i believe, by temperature and linearly? by current
<azonenberg_work>
(this is unrelated to the problem of parameter drift, temperature coefficient of resistance, etc making the circuit malfunction in a temporary fashion when too hot)
<azonenberg_work>
Eventually electromigation causes open circuit failures but this takes a long time
<azonenberg_work>
compare that to an arcing failure through gate oxide which is an instant short of gate to channel
<azonenberg_work>
Or Joule heating in conductors reaching the melting point
<azonenberg_work>
(yes i have seen this happen)
<pie_>
unfair amount of microscope time :pp
<azonenberg_work>
pie_: fair, i paid the same hourly rate as everyone else at the university
<azonenberg_work>
it just came out of my wallet instead of a grant
<pie_>
ok sorrry nevermind xD
<rqou>
are you _sure_ that's considered fair?
<azonenberg_work>
rqou: well sem time is hardly a basic human right
<rqou>
because when i asked at $FANCY_SCHOOL's cleanroom they said that the per-user fee was in addition to various university fees that ever lab had to pay
<azonenberg_work>
Policies may vary by school
<azonenberg_work>
at RPI, they had three rates
<azonenberg_work>
Internal users
<rqou>
and that these shared university fees made up the vast majority of the funding, hence why i couldn't get access at the student rate
<azonenberg_work>
Outside collaborating institutions (i.e. not paying RPI's standard lab overhead/rent etc fees)
<azonenberg_work>
And outside commercial users
<azonenberg_work>
at the time they were on the order of $45, $90, and $200 an hour
<azonenberg_work>
They didn't have a policy for students doing work outside of sponsored research
<rqou>
no monthly fee?
<azonenberg_work>
but after discussion with the lab manager, and agreeing to keep it non-commercial in nature
<azonenberg_work>
they set up an account for me at the internal-user rate
<azonenberg_work>
i dont think rpi had monthly fees for anything
<azonenberg_work>
they charged faculty overhead, but that was a percentage off the top of any grants they received
<azonenberg_work>
not time based
<azonenberg_work>
having a sem/fib account for a month without using the sem at all costs nothing
<rqou>
hmm in general it just sounds like rpi had less funding politics going on
<rqou>
maybe because it's a private school
<azonenberg_work>
yes, the politics were more along the lines of expelling student government members who dared to speak out against Her Highness
<azonenberg_work>
or abolishing the faculty senate when they did the same thing
<pie_>
lol
<azonenberg_work>
(sadly i am not making either of these up)
<azonenberg_work>
even tenured faculty were afraid to speak out against her publicly, but told me in private they hated her
<azonenberg_work>
its bad enough that a lot of alumni are now refusing to donate
<azonenberg_work>
So some higher-up, i think from the board of trustees? recently sent out an email accusing people of lying about their true motives
<azonenberg_work>
saying we actually weren't donating because she was a short black woman (or some subset thereof)
<azonenberg_work>
Here's a hint, if you squander your school's endowment, tank their credit rating, build a new performing arts center and football stadium while classroom buildings have leaky roofs and non-functional HVAC
<azonenberg_work>
and then call your alumni racists when they say you're a terrible president
<azonenberg_work>
it's not going to help your case :p
<pie_>
at least not with any sane people :P
<azonenberg_work>
another fun incident right after i graduated
<azonenberg_work>
There was a burglary in a dorm where somebody piggybacked on students going through the door
<azonenberg_work>
then went into a room and stole stuff
<pie_>
on the topic of magic smoke :P
<azonenberg_work>
so you know what their response was?
<rqou>
wait wait that activist woman is also an RPI alum?
<cr1901_modern>
rqou: Yes
<azonenberg_work>
They turned all but one door of all dorms into alarmed fire exits
<azonenberg_work>
And made it so you have to walk all the way around the building in a dark parking lot late at night to get to your room
<azonenberg_work>
instead of using the door facing the parking lot with a card reader on it :p
<pie_>
???
<azonenberg_work>
Yes, its an awful idea
<azonenberg_work>
Just like most things they did :p
<azonenberg_work>
They also built a tiny apartment inside each of the freshman dorms
<azonenberg_work>
So an assistant dean of students can helicopter-parent the students
<azonenberg_work>
Because apparently an RA isn't enough
<pie_>
eh. dont the british have something like that?
<pie_>
maybe im conflating it with boarding schools though
<pie_>
what an RA
<azonenberg_work>
resident... assistant? it's an upperclass student who lives in a dorm
<azonenberg_work>
does things like letting you into your room if your forget your keys
<azonenberg_work>
breaking up fights
<rqou>
dealing with people who got too drunk
<azonenberg_work>
yes, that
<azonenberg_work>
mostly that :p
<azonenberg_work>
But at least they're a student and not a member of the administration
<pie_>
oh
<pie_>
resident adult :P
<rqou>
lol
<rqou>
technically (almost) everyone is already an adult in the dorms
<rqou>
unless you mean a "real" adult? :P
<rqou>
oh hey since apparently this channel is ##openpolitics:
<rqou>
azonenberg_work: how do you feel about cody wilson? before and after the current arrest warrant
<pie_>
#thatsthejoke
* cr1901_modern
/part #openfpga
<rqou>
what? azonenberg_work is our resident gun nut :P
<azonenberg_work>
rqou: he's an idiot who is antagonizing the anti-gun crowd for no significant benefit
<azonenberg_work>
few if any of his products are viable replacements for either a COTS firearm, one machined in a proper machine shop, or a zipgun made out of plumbing supplies
<azonenberg_work>
all he's doing is getting more people mad and looking to ban stuff
<cr1901_modern>
azonenberg_work: Wow, azonenberg_work and I kinda-sorta agree on a gun-related topic ._.
<rqou>
yeah i pretty much agree with azonenberg_work as well
<cr1901_modern>
I couldn't care less if he uses an escort service. I heard that the minor was 13 years old (and not 16).
<azonenberg_work>
rqou: the current political climate is already anti-gun enough as it is, the last thing we need is to give them more reasons to crack down on things
<whitequark>
also isnt he like, a nazi
<rqou>
azonenberg_work: how would you feel about hypothetical things like fixing up that cad drawing whitequark and alt_kia(?) were discussing a while back?
* whitequark
facepalms
<whitequark>
i fixed it
<rqou>
oh you did already?
<whitequark>
then i deleted it because its dumb
<rqou>
lol
<rqou>
lol^2
<whitequark>
that was trivial to anyone with rudimentary mecheng knowledge and an ability to search youtube to see how it should look
<rqou>
but it's still illegal to make in 'murca
<rqou>
idk if the cad drawing would be ITAR-restricted though
<cr1901_modern>
whitequark: If he is a Nazi, I wasn't aware of that. Seems like a "typical" libertarian.
<whitequark>
cr1901_modern: "alt-right" or w/e
<cr1901_modern>
Oh, maybe...
<whitequark>
rqou: well not mine
<whitequark>
because im not subject to itar
<rqou>
lol right
<rqou>
cr1901_modern: founded h*treon
<whitequark>
could probably sell it via israel
<azonenberg_work>
rqou: what was it, ar auto sear?
<whitequark>
like the cryocoolers
<cr1901_modern>
rqou: Ahhh, well then ._.
<rqou>
azonenberg_work: iirc an even more hacky full-auto mechanical linkage thing
<whitequark>
azonenberg_work: the original drawing i converted to cad form was called "lightning link"
<azonenberg_work>
lol
<azonenberg_work>
oh
<rqou>
afaik analogous to the alleged ATF-serialized piece of string
<azonenberg_work>
See, i have no problem with the state of regulation on fully automatic firearms in the US
<whitequark>
i literally only did this because the drawing alt_kia posted had mutually conflicting dimensions called out
<azonenberg_work>
The tiny handful of civilian-legal ones that do exist are heavily restricted, you can get them if you live in the right state and have a few tens of thousands of dollars
<whitequark>
and it ground my gears
<rqou>
azonenberg_work: so what about serialized pieces of string?
<azonenberg_work>
But the last time one was used in a crime was like 40 years ago
<azonenberg_work>
so they're clearly restricted more than enough
<azonenberg_work>
on the flip side, i dont see much of a point in de-regulating them because there is little legitimate use for one
<whitequark>
"target practice"
<azonenberg_work>
Full auto is really only useful in a firefight with a military force that you are trying to suppress while another unit flanks them or while you break contact
<azonenberg_work>
And where you have a military supply chain to bring more ammo when you run out
<azonenberg_work>
it's useless for target practice because the accuracy is awful
<azonenberg_work>
It's useless for self-defense, the potential for collateral damage is absurdly high
<whitequark>
no i'm poking fun at the usual reason people say they need firearms
<whitequark>
reasons*
<azonenberg_work>
It's useless for hunting because a) high collateral damage potential and b) most hunting rules limit you to like a 5-round magazine or something
<azonenberg_work>
So even if you had full auto legally, it would be effectively useless given the mag size
<azonenberg_work>
also if you did manage to legally use it, put 30 rounds into a deer and there wont be much meat left :p
<azonenberg_work>
literally the only thing to do with one as a civilian is make noise at a range without actually gaining any experience that would help you shoot better in the future
<rqou>
azonenberg_work: so what about pieces of string? :P
<azonenberg_work>
rqou: Doesnt matter how you do it
<azonenberg_work>
bump firing, legal full auto, etc
<azonenberg_work>
uncontrolled rapid fire has no place in a civilian setting
<rqou>
sure, but isn't putting a serial number on a piece of string kinda silly?
<azonenberg_work>
It is
<azonenberg_work>
Although if i am ever in a public place getting shot at, i hope the guy has an actual machine gun with a ludicrous drum magazine
<azonenberg_work>
because a) most of his rounds will go into the ceiling and b) it will probably jam
<whitequark>
lool
<azonenberg_work>
So my odds of survival are higher than with aimed single shots from an ordinary mag
<azonenberg_work>
seriously, at ten rounds a second he'll run out of ammo pretty quick
<whitequark>
how many?!
<azonenberg_work>
600 RPM is a typical machine gun cyclic rate of fire
<azonenberg_work>
with a 30-round mag that's three seconds of shooting
<rqou>
wait really?
<azonenberg_work>
So maybe if you're the unlucky guy he points at first you get shot ten times instead of twice
<azonenberg_work>
but the five guys standing next to you are missed
<azonenberg_work>
and the ceiling is very screwed :p
<zkms>
thats why they invented 3 round burst or w/e
<zkms>
and also ammo belts
<rqou>
so other than full auto, how do you feel about silly restrictions on cad files?
<azonenberg_work>
rqou: what i DO object to is restrictions on suppressors
<azonenberg_work>
They dont make gunfire inaudible or even close
<rqou>
although imo it'd probably be a lot more useful to distribute ar lower cad files rather than silly 3d printed guns that can barely fire once
<azonenberg_work>
in most other countries with much more restrictive gun laws, they're highly encouraged or even mandatory safety equipment
<azonenberg_work>
like a muffler on a car
<azonenberg_work>
the US is unusually backward in that regard
<azonenberg_work>
whitequark, rqou: actually many are a lot higher
<azonenberg_work>
the ak47 is 600 RPM, the M16A1 is 700-950
<zkms>
yah suppressors in the NFA is stupid; like, it's trivial to make an improvised suppressor with an oil can and a thread adaptor and yet there's barely *any* crime (beyond like...non-permitted possession of a suppressor) committed with improvised suppressors (because they make handguns far less concealable and most gun crime in US is done with illicitly-carried handguns)
<azonenberg_work>
The M249 is adjustable 725 or 1000
<zkms>
oil filter not oil can lol
<zkms>
i should go to sleep ;;
<azonenberg_work>
zkms: exactly
<rqou>
azonenberg_work: so how would you feel about a not-alt-right person fighting the government to distribute AR lower cad drawings rather than a 3d printed piece of crap?
<zkms>
rqou: that's a moot case, CAD files for AR15 lowers have been online for many many years now
<azonenberg_work>
exactly
<rqou>
oh really?
<azonenberg_work>
people have been DIYing AR lowers for years
<zkms>
^
<rqou>
yeah i know that
<azonenberg_work>
They have companies that sell 80% lowers (i.e. just barely not a firearm according to ATF rulings)
<zkms>
people have made AR lowers out of fuckin wood lol
<azonenberg_work>
that just needs a little milling and drilling to be functional
<rqou>
but is there like a gcode that you can just download and stick in a mill?
<azonenberg_work>
i think gcode is normally very machine specifi
<azonenberg_work>
specific*
<azonenberg_work>
as far as tool selection, feed speeds, etc
<azonenberg_work>
you can almost certianly get a STEP that you can do your own CAM work for your machine with
<rqou>
hmm yeah that's true
<azonenberg_work>
but honestly, 99% of people who want an AR just buy either a whole one, or a stripped lower and furnish it themselves
<rqou>
oh lol you really can instantly find such a file on google
<azonenberg_work>
there's just a handful of extremists who are worried obama is gonna come and confiscate everything and want something off the record
<rqou>
as a STEP file
<whitequark>
yeah but you dont want step
<whitequark>
or gcode
<azonenberg_work>
Despite the fact that if you did a union of the customer lists for (say) cheaperthandirt, midwayusa, and one or two other big shooting-sports retailers
<whitequark>
instructions for a manual mill would be more convenient
<rqou>
it has an Inventor drawing too
<azonenberg_work>
you could probably find 98% of US gun owners in a matter of hours with a handful of subpoenas
<azonenberg_work>
Even if they have never bought a gun from a dealer and thus never been through a NICS check
<azonenberg_work>
so the lack of a national registry would do almost nothing to stop an actual totalitarian government from confiscating them if they really wanted to
<azonenberg_work>
(which is the main argument for unserialized lowers, that nobody knows you have them)
<rqou>
hmm this basic shape is so simple it probably actually can be made on a manual mill
<whitequark>
???
<whitequark>
how do you think people made firearms before CNC
<pie_>
caught you <zkms> oil filter not oil can lol
<whitequark>
or like... anything
<azonenberg_work>
rqou: people go from 80% to finished with manual mills all the time
<azonenberg_work>
i've heard of it beoing done with a drill press
<whitequark>
you dont need shit like 5 axis mills for 99% of work
<rqou>
yeah i know that
<azonenberg_work>
a more skiled machinist would have no problem doing the whole lower from bar stock
<rqou>
huh the ar-15 design is _old_
<zkms>
yah
<rqou>
i thought it was much more complicated than it actually is lol
<zkms>
semi-automatic rifles have been available for a looong time
<azonenberg_work>
rqou: you've seen mine naked, it's not all that complex
<pie_>
i was actually mildly interested in gunsmithing at one point, still am I guess, but no environment for it
<whitequark>
now if you made an fn f2000 on a manual mill
<whitequark>
thatd be impressive
<rqou>
the shape of commercial lowers is somewhat complicated-looking
<rqou>
but i guess you can make one that's simplified but still has all the critical parts
<whitequark>
does it have go faster bumps
<rqou>
the cad file i'm looking at has chamfers and stuff :P
<pie_>
i googled f2000 internals but i only get airsoft stuff :/
<rqou>
but you don't really need those
<azonenberg_work>
rqou: the only chamfer that matters for functionality i think is the one on the mag well
<azonenberg_work>
So it goes in smoothly without snagging
<whitequark>
you can just make chamfers on a manual mill you know
<whitequark>
with minimal setup
<azonenberg_work>
And that
<azonenberg_work>
yeah its just an angled cutting bit
<azonenberg_work>
Or a tilted workpiece and ordinary endmill
<rqou>
but this is a _rounded_ chamfer :P
<pie_>
i actually dont quite understand why theres so many variations on guns
<rqou>
so i guess it's technically a fillet?
<pie_>
but thats just me being dumb
<whitequark>
pie_: f2000 is mostly made from polymers and composites iirc
<rqou>
anyways, if clickspring videos are any indication, anything that is "hard" to make on a mill can probably be made to work by hand filing it
<azonenberg_work>
pie_: there really arent THAT many variations, just subtle improvements or switching from all metal to more polymer construction
<azonenberg_work>
a lot of modern designs look very similar under the hood
<azonenberg_work>
there's a trend toward more modularity and lighter, polymer designs with only critical parts made out of metal
<pie_>
aha...
<whitequark>
rqou: LOL
<pie_>
somewhat related, are old-style machinists that can do stuff with their hands dying out? is it importsnt?
<whitequark>
nope
<azonenberg_work>
but no earth-shattering engineering improvements i can think of recently
<azonenberg_work>
pie_: i know quite a few of them, they're out there
<pie_>
oh ok
<whitequark>
manual and cnc work isnt inherently different
<azonenberg_work>
For a one-off, a skilled manual operator may get it done faster than writing gcode and configuring a CNC
<rqou>
whitequark: why lol?
<whitequark>
there's a decent amount of skill transfer between those
<zkms>
caseless ammo!1!!1!
<whitequark>
for a one-off manual operator WILL get it faster
<whitequark>
most of the time
<azonenberg_work>
unless it has really weird curves
<azonenberg_work>
CNC is great at strange shapes
<zkms>
i'm going to sleep i have a 0600 flight to catch tomorrow and it's already 0015
<whitequark>
CNC is great if you need 1000 of them
<pie_>
zkms, nth attempt ;P
<whitequark>
though people did that before CNC
<whitequark>
in lke 1800s
<whitequark>
with cam shafts
<rqou>
ooh right
<azonenberg_work>
whitequark: or just making jigs to align things
<zkms>
good night y'all ~
<whitequark>
yep that too
<pie_>
zkms, just close the computer, youll have better luck xD
<azonenberg_work>
so you could put th eworkpiece in the jig and just swing the mill left and right until it hit the end stop of the jig
<azonenberg_work>
etc
<whitequark>
any really skilled cnc programmer will have to understand most of what a manual mill operator needs to know
<whitequark>
3axis cnc needs a ton of setup skills
<pie_>
i guess im just frustratingly removed from anyone that can actually make anything
<rqou>
heh at $FANCY_SCHOOL only the absolute most trusted students were allowed to use the CNC
<whitequark>
go to russia lol
<pie_>
whitequark, yeh
<whitequark>
im serious
<rqou>
with machinists staring at you/the machine the whole time in case you break something
<whitequark>
machinist work is super easy here
<pie_>
i kind of wanted to apprentice at the local university glassblower but he didnt really seem into it
<pie_>
*scientific glassblower
<azonenberg_work>
pie_: there's a small job shop like a mile from the place i used to rent
<azonenberg_work>
maybe a quarter mile from lain and monochroma's place
<pie_>
:P
<azonenberg_work>
i had them do a little project for me a little while back
<azonenberg_work>
the helmet mount for my thermal imager didn't have quite enough adjustment range to fit my eyes
<azonenberg_work>
So i had him extend the groove a little bit
scrts has quit [Ping timeout: 246 seconds]
<rqou>
i wish i had the space/money to have a mill in my workspace
<azonenberg_work>
Me too, but a little mini PnP is the higher priority for me
<azonenberg_work>
As that is something i could actually use on a frequent basis
<azonenberg_work>
a laser cutter would be nice for enclosure faceplates
<azonenberg_work>
but i design 3D items (vs 2D faceplates) rarely enough that i think i'm better off hiring that out
<azonenberg_work>
it'd take up too much space in my lab
<rqou>
oh right thanks for reminding me to get my fucking k40 upgraded
<rqou>
hmm i wonder what's better: owning a mill+lathe or making friends with a machinist?
<whitequark>
k40?
GuzTech has joined ##openfpga
m4ssi has joined ##openfpga
<whitequark>
rqou: latter
<zkms>
^
<rqou>
those "china-special" "40 watt" co2 laser cutters
<whitequark>
a skilled machinist is basically a metal magician
scrts has joined ##openfpga
<whitequark>
also, welders
<whitequark>
i know a guy who can weld 0.3mm 316L by hand
<whitequark>
this is basically, foil
<rqou>
btw hilarious thing about those chinese laser cutters
<rqou>
i was telling my sister about them and explaining how you normally control them using a shady plugin for a pirated version of coreldraw on a pirated windows xp (with DEP/NX off of course)
<rqou>
and then my sister just says "oh yeah, we've got one of those in our research lab"
<whitequark>
lol
<rqou>
apparently one of the grad students added an e-stop safety switch :P
<rqou>
oh yeah, by default these don't have any interlocks whatsoever
<rqou>
so really good for causing "remaining eye" scenarios
<rqou>
oh btw did i mention that this piece of crap software requires a dongle?
<whitequark>
azonenberg_work: oh thats easy
<whitequark>
have a STEP file?
<azonenberg_work>
I want something that fits around a SMA connector and lets me tighten/loosen it from a standard hex drive socket
<azonenberg_work>
using a torque wrench/driver
<whitequark>
or do you want a design?
<azonenberg_work>
I dont have a design yet but could probably throw one together in solvespace fairly quickly
<rqou>
oh and the reason it requires DEP off? it's packed with a shitty packer
<rqou>
yes, chinese crapware with drm
<azonenberg_work>
any idea what it'd cost? preferably made out of 316 SS or maybe aluminum (not sure if alu would be strong enough, probably given that i plan to run below 10 in-lb torque)
<whitequark>
azonenberg_work: can't quote machining without a STEP file or drawings
<azonenberg_work>
i'll bang one up in solvespace when i get a chance then
<whitequark>
well probably less than $200 :P
<azonenberg_work>
and lol
<azonenberg_work>
that was the question :p
<zkms>
i know someone who posted something on twitter about a 3d printed SMA wrench in metal
<azonenberg_work>
i.e. doing it from scratch is probably cheaper tha buying thers
<whitequark>
usually it goes like this, i give the shop a STEP file, they do drawings according to ГОСТ for a small fee, then machine according to drawings
<rqou>
what happens when a clueless customer gives them something un-machinable?
<whitequark>
i tried making drawings myself but they hated them
<whitequark>
rqou: they tell you
<whitequark>
this isnt like, an official shop
<whitequark>
last time i was making something, i had a contact pass drawings to some guys moonlighting on a closed object
<azonenberg_work>
lol
<whitequark>
they made absolutely outstanding high vacuum hardware for close to nothing
<whitequark>
i havent even asked for cost breakdowns because it was so laughably cheap
<whitequark>
rqou: although an official shop would also tell you
<whitequark>
its pretty easy to get a flat out refusal
<whitequark>
if you appear clueless
<whitequark>
guess how i know :p
<pie_>
whitequark, i find you appearing clueless hard to believe :P
<whitequark>
pie_: hell no
<pie_>
of course appearances can be deceiving
<whitequark>
im clueless all the time
<pie_>
;P
<whitequark>
because i constantly try to get into completely new areas
<rqou>
whitequark: how are people like you and @profanegeometry so good at finding these local businesses?
<pie_>
i dont try, I do it by accident lol
<whitequark>
rqou: no idea how she does it
<rqou>
am i just terrible at people skills?
<whitequark>
i just go on like, craigslist
<whitequark>
and call
<pie_>
rqou, i think its russia
<whitequark>
and then keep numbers
<pie_>
or hong kong
<pie_>
idk
<rqou>
ugh but then i need to use the _phone_ :P
<pie_>
what oyu dont have a shady noir trenchcoat to show up at people's doors in?
<rqou>
seriously, i absolutely hate making/receiving phone calls
<whitequark>
sure but i hate not getting parts machined more
ZipCPU|Laptop has quit [Ping timeout: 272 seconds]
<pie_>
<pie_>
what the fuck two bigass sonic booms just a thing
<rqou>
also afaict in the US random shady machinists don't really advertise on craigslist
<whitequark>
machining in the us is way too epxensive anyway
mumptai_ has quit [Remote host closed the connection]
<pie_>
rqou, ok maybe the problem isnt even that im in hungary there is just literally nothing in this ciry
<pie_>
city
<rqou>
pie_: move to russia or hong kong? :P
<pie_>
which is probably not true, but thats what it feels like
<rqou>
or sillycon valley? :P
<pie_>
$hard :P
<pie_>
rqou, silicon valley works fo me if you give me a corner of your room and pay for food :p
<rqou>
lol
<rqou>
i don't mind, but i don't think my parents would be happy
<pie_>
im very nice
<whitequark>
lol
<rqou>
yeah, but my parents aren't millennial and are chinese :P
<rqou>
so "somebody from the internet living in a corner" doesn't fly :P
<jn__>
how about "somebody from hungary living in a corner"? :D
<rqou>
lol
<pie_>
how does that help?? xD
<pie_>
rqou, maybe you can list more dependents for tax purposes :P
<whitequark>
lol
<rqou>
lool
<azonenberg_work>
whitequark: can you PM me your email address again? wanted to send you a draft SLVS for the tool
<rqou>
wtf how do you have time to work on this?
<whitequark>
whitequark@whitequark.org :p
<azonenberg_work>
rqou: i threw it together in 20 minutes while waiting on a compile
<azonenberg_work>
:p
<azonenberg_work>
its not complex
<azonenberg_work>
its a positive and negative hex extrusion
<azonenberg_work>
two metal plates
<azonenberg_work>
and a bar holding them together
<azonenberg_work>
whitequark: sent, have a look?
<whitequark>
azonenberg_work: that email signature
<azonenberg_work>
whitequark: HR made us do it
<rqou>
oh loool
<azonenberg_work>
i have no choice in the matter
<rqou>
did you use your ioa email?
<azonenberg_work>
Yeah i'm on the laptop
<rqou>
btw pie_ i just checked and you can't claim a random person as a dependant :P
<rqou>
they have to be related to you somehow
<whitequark>
marry him
<rqou>
loool
<rqou>
i think that actually would work lol
<rqou>
##opentaxevasion :P
<pie_>
lmao
<pie_>
whitequark, i was surprised it took that long for that suggestion to come up
<whitequark>
azonenberg_work: lgtm, sending it for a quote
<pie_>
alternatively, we all knew where this was going
<rqou>
hmm actually maybe not
<rqou>
it's unclear if "random people who live with you" can count as a household or not
<whitequark>
azonenberg_work: do you have any critical dimensions
<rqou>
definitely inviting audits if you try lol
<whitequark>
or do normal machining tolerances do fine
<azonenberg_work>
Normal tolerances are fine, this is not a 100% final file
<whitequark>
ok
<azonenberg_work>
i need to double check actual measurements on real mating parts with calipers
<azonenberg_work>
I will build some slop into the final dimensions
<azonenberg_work>
maybe 25-50 um
<azonenberg_work>
for tolerance of both this tool and the mating part
<azonenberg_work>
It's a wrench so being a little loose is OK, as soon as you turn it will make contact
<rqou>
damn irs rules are pretty awful with how they interact with "family"-related laws
<azonenberg_work>
rqou: now you see why the gays want to get married so badly?
<azonenberg_work>
marrying $wife was a massive tax break for me, she made almost nothing but it moved me down a tax bracket
<rqou>
i mean, it depends
<rqou>
usually it's only beneficial if there is a large difference in incomes
<whitequark>
azonenberg_work: sent
<azonenberg_work>
rqou: yeah, which there was here
<whitequark>
azonenberg_work: is this like a rush thing or?
<whitequark>
the usual lead time is 1-2 weeks but it's probably possible to do it faster
<pie_>
the job or the marriage?
<whitequark>
"the gays" lol
<rqou>
hmm family law itself is a nice giant pile of hacks
<rqou>
thanks christianity
<pie_>
rqou, im guessing something about balancing things and rich people
<pie_>
...that was not one of my more descriptive sentences
<pie_>
ok im off...gonna try to get myself to figure out how to do this data structure i need...
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
<azonenberg_work>
whitequark: far from a rush, i dont have a lab to use it in yet
<azonenberg_work>
what happened is, i bought a torque screwdriver for electrical work
<azonenberg_work>
Because circuit breakers and receptacles/switches all specify a tolerance range for torque to get a solid connection that wont loosen easily
<azonenberg_work>
And it seemed only natural to properly torque my SMA connectors going forward
<azonenberg_work>
I have a T-handle SMA wrench preset at 8 in-lb but thats too much for a lot of the brass connectors
<azonenberg_work>
My screwdriver is variable
<azonenberg_work>
So I needed an adapter to go from my 1/4" hex drive screwdriver to a SMA
<rqou>
what torque screwdriver do you have?
<whitequark>
oh yeah i just remembered this discussion we had before
<azonenberg_work>
whitequark: basically time frame is "before i move into the new lab" lol
<azonenberg_work>
right now we are looking at that happening circa end of this year to early january
<whitequark>
azonenberg_work: so a few years right
<azonenberg_work>
rqou: one from the home depot, not a super fancy one but its adjustable in single digit in-lb
<rqou>
i agree with whitequark :P
<whitequark>
>in-lb
<whitequark>
please
<azonenberg_work>
?
<rqou>
azonenberg_work: does it mechanically disengage or just beep at you?
<azonenberg_work>
mechanically disengages
<whitequark>
azonenberg_work: units :P
<azonenberg_work>
it starts clicking with some kind of ratchet and stops moving the screw
<rqou>
is it accurate? have you calibrated it?
<azonenberg_work>
It came with a calibration certificate saying it was +/- 6% from the factory
<rqou>
hmm
<azonenberg_work>
Have not recalibrated since but that was only like a month ago
<rqou>
meh
<rqou>
not super accurate
<azonenberg_work>
well its a $50 torque driver, not a $500 one
<rqou>
ahh ok
<azonenberg_work>
That's good enough for my needs, the receptacles want 12-14 in-lb on the terminals
<azonenberg_work>
If i set for 13 i'm within the range at +/- 6%
* jn__
should build something that uses metric fake inches (25mm) and metric pounds (0.5kg)
<azonenberg_work>
jn__: i do that on my PCBs lol
<pie_>
...metric fake inches?
<azonenberg_work>
the official design rule at oshpark is in mils, 5 mil which is like 127 um or something
<rqou>
inb4 somebody talks about chinese customary units
<whitequark>
jn__: youre evil
<azonenberg_work>
I use 125 um as my design rule
<azonenberg_work>
nobody's complained yet and it's cleaner in metric
<azonenberg_work>
i dont expect yield issues from being 2um below the advertised limit :p
<jn__>
i once tried to use a 0.1" pitch DIP chip on a 2.5mm pitch perfboard... it didn't work well.
<pie_>
you have to bend the legs more
<whitequark>
related: i hate "nominal dimensions"
<rqou>
jn__: btw PRC 斤 ("pounds") are exactly the same as metric pounds
<rqou>
but HK pounds are ~100g heavier
rohitksingh_wor1 has joined ##openfpga
<rqou>
well, HK 斤
<rqou>
because british units occasionally get used too
<whitequark>
what is a dimension (any dimension) of a 1" nominal diameter pipe? no one can tell but sure as heck not 1"
rohitksingh_work has quit [Ping timeout: 245 seconds]
<whitequark>
rqou: wtf
<whitequark>
no
<jn__>
rqou: german pounds are metric, too (but i don't know how much they were before the kilogram was invented)
<rqou>
whitequark: they have different names in chinese
<whitequark>
rqou: what
<rqou>
ok wait let me double-check
<azonenberg_work>
whitequark: electrical conduits are no better
<rqou>
ok, kilogram = 公斤; customary unit 市斤 is 500g in PRC, ~605g in HK, and is traditionally called "catty" in english; british pound = 磅
<azonenberg_work>
look at the ID and OD of a 3/4 inch trade size conduit
<azonenberg_work>
neither is 0.75 inch :p
<rqou>
so yeah, i mixed up what 斤 is normally translated to in english
<rqou>
oh wtf
<rqou>
there's a wikipedia page that calls 斤 "Chinese pound"
<rqou>
since no normal person uses "catty"
<azonenberg_work>
rqou: what about short tons
<azonenberg_work>
long tons
<azonenberg_work>
and metric tons? :p
<whitequark>
>The catty is traditionally equivalent to around 1⅓ pound avoirdupois
<whitequark>
gah
<whitequark>
>In mainland China, the catty (more commonly translated as jin within China) has been rounded to 500 grams and is referred to as the market catty (市斤 shìjīn) in order to distinguish it from the "metric catty" (公斤 gōngjīn), or kilogram, and it is subdivided into 10 taels rather than the usual 16.
<rqou>
no, i don't know how to actually use any of these
<rqou>
although apparently mechanical engineers still use them
<rqou>
azonenberg_work: also, how do you feel about measuring capacitance in centimeters? :P
s_frit has quit [Read error: Connection reset by peer]
s_frit has joined ##openfpga
s_frit has quit [Remote host closed the connection]
s_frit has joined ##openfpga
<rqou>
whitequark: apparently TW units can be even more cursed
<rqou>
> Taiwanese units of land measurement derive from both traditional Dutch and Japanese measurements. The principal unit of land measure, the jia (甲; jiǎ; kah), derives from the obsolete Dutch morgen, which was introduced during Taiwan's Dutch era. The li (犁; lí; lê) represented the area that could be farmed by one man with one ox and one plow in one day.
<gruetzkopf>
ah, obsolete european area units
<gruetzkopf>
the morgen was extremely dependant on location, as the amount of work per area required varies greatly
<rqou>
lol
<rqou>
weird legacy units in all sorts of places
<rqou>
hey whitequark, how do you feel about the use of the пуд for measuring weird sports weights?
m_t has joined ##openfpga
rohitksingh_wor1 has quit [Ping timeout: 246 seconds]
<felix_>
rqou: regarding the discussion yesterday: another area where responsible disclosure imho is the best thing to do are bugs that lead to unauthorised access of sensitive user data. at least i wouldn’t want to possibly endanger users only because some company didn’t do their security-related homework
genii has joined ##openfpga
Bike has joined ##openfpga
Miyu has joined ##openfpga
rohitksingh has joined ##openfpga
Prf_Jakob has quit [Changing host]
Prf_Jakob has joined ##openfpga
<whitequark>
rqou: uh
<whitequark>
does anyone actually do that?
<whitequark>
azonenberg_work: yeah I specified 316L
<felix_>
whitequark: are you still interested in the thunderbolt cc-line trace when plugging in the tb16-dock?
<whitequark>
felix_: yep!
<felix_>
ok, then i'll try that later
<whitequark>
thanks
<awygle>
Pounds are supposed to be force but really aren't. I hate nominal dimensions even more than typical datasheet values
mietek has joined ##openfpga
mietek has joined ##openfpga
mietek has quit [Changing host]
<Bike>
oughta use slugs... which textbooks are actually pretty good about in my experience, at least
ayjay_t has quit [Read error: Connection reset by peer]
ayjay_t has joined ##openfpga
<shapr>
slugs teach FPGA? what?
ZipCPU|Laptop has joined ##openfpga
<Bike>
slugs are the proper unit of mass in whatever the units system the US uses is called
<cpresser>
i am building a modbus (rs485) device and am wondering which kind of pretection I need for which use-case?
<cpresser>
are tvs diodes required if I already have clamps?
ZipCPU|Laptop has quit [Ping timeout: 246 seconds]
mietek has left ##openfpga ["NO CARRIER"]
<qu1j0t3>
awygle: did you fix the oscillation?
<awygle>
qu1j0t3: no, I went to bed.
GuzTech has quit [Quit: Leaving]
wpwrak has quit [Ping timeout: 245 seconds]
wpwrak has joined ##openfpga
wpwrak has quit [Ping timeout: 252 seconds]
<azonenberg_work>
cpresser: there is a subtle difference
<azonenberg_work>
if memory serves me right, clamping diodes are zeners from signal to ground that break down at a specified voltage
<azonenberg_work>
while steering diodes are diodes from signal to power and ground that keep you between the rails?
<azonenberg_work>
could be wrong on the terminology but both are commonly used for protection
<awygle>
cpresser: tvs diodes are often much faster than clamps, so often you do need both
<whitequark>
awygle: azonenberg_work: how does ESD protection in ICs actually work
<whitequark>
like i know it's diodes to power rails
ZipCPU|Laptop has joined ##openfpga
<whitequark>
but WHY does this protect from ESD
<whitequark>
as I understand CMOS inputs are ESD-sensitive because in their unaltered state they're high-impedance
<azonenberg_work>
whitequark: when the voltage at the pad gets more than a diode drop above Vdd
<whitequark>
and so any charge that builds up on them will eventually just break down the dielectric
<azonenberg_work>
the diode starts conducting and quickly shunts the charge into the power rail (or ground, depending on the voltage)
<azonenberg_work>
They're taking advantage of the fact that the ESD pulse has a high voltage but a very limited total energy
<whitequark>
right, but why can't there be for example
<azonenberg_work>
So if you discharge it fast enough it wont do much damage
<whitequark>
a ESD pulse between power rails with reverse polarity
<whitequark>
why are these diodes *sufficient* for any sort of ESD no matter where
<azonenberg_work>
very often there are also zeners between the power rails
<azonenberg_work>
for exactly that reason
<azonenberg_work>
Also, power rails tend to have lots of capacitance
<azonenberg_work>
Given the limited total charge in an ESD pulse, the voltage will dissipate pretty quickly
<azonenberg_work>
The human body model for ESD resistance is 100 pF at the specified voltage in series with 1.5 kohms
<azonenberg_work>
100 pF at a few kV will easily put dangerous voltage on a 2 pF input buffer
<awygle>
Right. Power rails are basically modeled as infinite charge sinks here, because you get a couple nC of actual charge so relative to many uF on power rails it's not a big deal
<azonenberg_work>
But a power rail with multiple uF of capacitance, or even a nF or so of on-die power plane capacitance, will barely move
<azonenberg_work>
You can kill a die with sustained overvoltage easily but ESD tends to have a very limited duration due to the capacitance of a person
<azonenberg_work>
Incidentally, this is what made the starshipraider protection circuit so complex
wpwrak has joined ##openfpga
<azonenberg_work>
Since i was trying for resistance to DC faults, and you can't rely on the limited charge of the ESD source
<awygle>
Yep. Sustained anything is a much bigger problem. "esd protection is like primary batteries" -Michael Scott
<azonenberg_work>
So even though i wanted +/- 12V sustained vs +/- 8 kV ESD pulses
<azonenberg_work>
the DC fault was much more challenging to protect against
ZipCPU|Laptop has quit [Ping timeout: 252 seconds]
<whitequark>
ooooooh
<whitequark>
I see now
<whitequark>
my favorite kind of question, "how the hell this does work in real life without X?" "it uses X"
<awygle>
Haha
<azonenberg_work>
an ESD diode can be fairly small since, although it only has to handle high fault current for a couple nanoseconds
<azonenberg_work>
it can rely on thermal inertia of the surrounding die area to sink heat
<azonenberg_work>
So you can use a fairly small, low capacitance diode
<azonenberg_work>
For starshipraider i needed to use a large, beefy diode that can sustain the faults for an indefinite time period
<whitequark>
why did early CMOS not include ESD diodes?
<azonenberg_work>
Which meant a huge diode, which meant high parasitics that destroy signal integrity
<awygle>
low cap esd diodes are super interesting
<awygle>
PNNP ftw
<azonenberg_work>
whitequark: i think they did, they were just much less effective than the modern ones?
<awygle>
Or is it nppn...
<whitequark>
oh
<whitequark>
why were they worse?
<azonenberg_work>
probably higher R(on), slower reaction time, etc
<awygle>
good diodes are hard to make
<awygle>
just like good transistors
<whitequark>
huh
<azonenberg_work>
i mean VERY early stuff didnt inclue it becuase it probably would have taken too much die area
<whitequark>
I think you could kill ISA hardware super easily with ESD
<whitequark>
and I can probably rub a modern PCI-e card on the cat and nothing will happen
<awygle>
Aren't most on die diodes basically diode wired transistors and super non optimized? Or am I thinking of something else
m4ssi has quit [Remote host closed the connection]
<rqou>
> A 19-year-old man from the United Kingdom who headed a cybercriminal group whose motto was “Feds Can’t Touch Us” pleaded guilty this week to making bomb threats against thousands of schools.
Bike has quit [Ping timeout: 256 seconds]
indy has joined ##openfpga
<felix_>
whitequark: http://sigsegv.notmysegfault.de/intern/xps13-tb16.sr the usb pd decoder didn't want to decode that, but i'd say that this trace doesn't look as wrong as the one i got without the comparator between the cc line and the logic analyzer
<felix_>
captured with 2mhz and i adjusted the other voltage that goes into the comparator so i see the signal and get clean edges
<rqou>
what the fuck. apple apparently purposely disables DP MST daisy chaining
<rqou>
whitequark can you hack this for me?
<gruetzkopf>
macos has no MST support at all
prpplague has quit [Quit: Leaving]
ZipCPU|Laptop has joined ##openfpga
<rqou>
apparently that's not completely true
<rqou>
first of all, the hardware does MST just fine and apparently windows can use it
<rqou>
also, apparently the MST "two panels duct-taped together" 4k/5k displays are also supported
<rqou>
so apple just artificially limited MST daisy chaining
Bike has joined ##openfpga
wpwrak has quit [Read error: Connection reset by peer]
wpwrak has joined ##openfpga
wpwrak has quit [Ping timeout: 252 seconds]
wpwrak has joined ##openfpga
<awygle>
yo Security People, are containers on bare metal (as opposed to on VMs) horribly insecure in Linux-land?
<jn__>
privesc from a container to the OS shouldn't be easier on bare-metal than inside a VM
<jn__>
privesc from container to OS vs privesc from VM guest to VM host is a different question
<awygle>
i have a hard time seeing how "add a VM layer" could make it _less_ secure. but i know freebsd jails and solaris zones are considered "pretty secure generally", and i know at one time it was considered really easy to escape from a container in linux
<awygle>
i only care a little bit about security but if i did start running a public web facing service or something i'd want to at least take minimal steps, but VMs seem pretty heavyweight and containers are an appealing alternative
<awygle>
or maybe instead of a "reinvent the OS' existing isolation mechanisms" technology i should just run everything as not-root, lock the permissions down, maybe run selinux, and call it a day. i really don't know how to evaluate those choices
<rqou>
i thought most of the escapes have been fixed?
<awygle>
that's basically my question. i'd be happy to hear that that was true
<rqou>
user namespaces introduced a nice pile of new vulns unfortunately
<rqou>
also docker managed to introduce a nice set of new potential misconfigurations
<awygle>
huh, user namespaces seem very cool, but also i can definitely see how "whoops you're actually root _everywhere_" could happen pretty easily
<rqou>
tl;dr in my personal opinion you don't necessarily gain much security against determined attackers using containers
<rqou>
but you gain some protection against stuff like "oops XXE" or "oops LFI"
<rqou>
and you "simply" (or just pick different tradeoffs) configuration/change management
<rqou>
*simplify
Miyu has quit [Ping timeout: 252 seconds]
<awygle>
"gain much security" compared to what, just unprivileged processes?
<rqou>
yeah
<awygle>
mk
m_t has quit [Read error: Connection reset by peer]
<rqou>
note that i haven't actually tried this, but they looked legit last time i looked into this
<awygle>
this is a service to intercept spam calls for you?
<rqou>
no, to make spam calls :P
<awygle>
oh okay. i knew it was one or the other lol
<awygle>
in that case what the _fuck_ does "Anveo Direct service will soon offer wholesale E911 service in US/Canada." mean?!
<gruetzkopf>
DID is more fun on a non-sip line in the german numbering plan
<gruetzkopf>
that also does inbound
<rqou>
oh yeah, if you go through the "wholesale" side e911 is unbundled
azonenberg_work has quit [Ping timeout: 245 seconds]
<rqou>
because there are fees attached :P
<rqou>
anyways, these guys seem to be much more legit and not actually intended for you to make spam calls
<awygle>
i clearly understand nothing about how phones work because this is not making sense to me
<rqou>
they're really popular on dslreports
<rqou>
oh yeah, this industry is bonkers
<rqou>
this wholesaler completely unbundles calling other people, received calls on a number, and now e911
<rqou>
they're a reseller of service from "real" telcos
<rqou>
afaict they do the call setup stuff (translating to/from SIP?) and then help you forward pcm sampled audio into the telco networks
<awygle>
i don't understand why there's a wholesaler involved in _emergency services_ at all
<rqou>
i don't quite get it either
<rqou>
it's an artifact of the US billing environment
<gruetzkopf>
my current SIP provider (sipgate) for $MYCORP location signals location to emergency services as well
<rqou>
in the us, voip that is "like a real phone plan" (e.g. Vonage) are required to do so
<rqou>
they also collect some kind of fcc fee that is used to subsidize other stuff
<rqou>
wholesalers and people who skirt the rules (e.g. GVoice) don't send 911 info and don't pay the fee
<rqou>
<insert joke here about how the telephone network is a billing system that happens to be able to exchange sound information>
<gruetzkopf>
i pay my provider by possible number of concurrent calls inbound, by number of extensions and by outbound call time
<rqou>
that sounds like a normal "retail" plan
<gruetzkopf>
sip doesn't like overlap dial though
<awygle>
i did realize the other day that i probably should get at least a basic land line
<rqou>
meh
<awygle>
turns out if you leave your cell phone in a friend's car it's nice to be able to call them
<rqou>
my parents finally cancelled our POTS connection
<emeb>
or just get another cellphone.
<rqou>
at&t kept price gouging more and more to try to get people to cancel
<emeb>
you can get a cheap pay-as-you-go backup for like $30/yr
<rqou>
like $20/mo for a basic landline with no long distance
<gruetzkopf>
so it's far less flexible than ISDN or IKZ-50 analog DID
<rqou>
nobody does that in the US
<rqou>
the US telecommunications ecosystem _sucks_
<gruetzkopf>
with SIP, you need to pre-set each callable number with your provider
<rqou>
afaik no?
<gruetzkopf>
i provisioned -0 and -10 to -90 for $MYCORP-LOCATION (yes, the latter are 1 digit longer
<rqou>
not with the wholesaler?
<rqou>
oh, for other people to call in?
<gruetzkopf>
yeah
<rqou>
yeah, i bet real ss7 access allows you to do a hijack
<gruetzkopf>
with ISDN and the older analog IKZ-50 scheme from the SxS and rotary dial exchanges, your PBX got all digits A party dialed as soon as the prefix was reached and until the PBX signaled "ringing" back to a
<rqou>
hey gruetzkopf, do you have any good resources for me to learn about BGP?
<gruetzkopf>
and it's not "reach dialtone, dial again" bullshit which incurs costs to A
<rqou>
oooh that use case
<rqou>
is that even possible with how call setup gets done nowadays?
<gruetzkopf>
SS7 was designed with it in mind
<gruetzkopf>
because everyone did it
<gruetzkopf>
at least in europe
<gruetzkopf>
H.323 is also not totally broken in that respect
<lain>
[off-topic] anyone know what determines cmake's search order for find_package? I have two arch machines, both up to date, cmake 3.12.2, building something that uses libxml2. one machine succeeds, the other fails due to wrong include dir for libxml2. I am struggling to determine why the two behave differently... libxml2-config.cmake is mildly broken, it uses ${CMAKE_CURRENT_LIST_DIR}/../../../include as the
<lain>
include dir, which works if it's in, say, /usr/lib64/cmake/libxml2/libxml2-config.cmake, giving /usr/include. what happens on the failing machine is it searches /lib64 (symlink to /usr/lib) first, finds /lib64/cmake/libxml2/libxml2-config.cmake, then dutifully assigns an include dir of //include, which is of course invalid...
<lain>
trying to figure out why cmake is searching /lib64 first on one system, but /usr first on the other
<rqou>
ping whitequark?
<gruetzkopf>
so for overlap-dial telcos know your prefix-length, and once that is reached, you get SETUP and then INFO per digit dialed until your PBX says "yep, ringing" "no busy" "no, not existing" or one of the other ISDN error codes
<gruetzkopf>
but those 3 existed in pure-analog times
<lain>
the actual search order on the failing machine is /, /lib64/cmake. on the successful machine it is /usr/local, /usr, /usr/lib64/cmake
<gruetzkopf>
"exists, not busy, but no path in time/space/memory crossbar available" is a seperate error code :D
<rqou>
lol
<rqou>
nowadays i don't really consider telephone stuff all that relevant
<gruetzkopf>
i really like what they envisioned with ISDN
<rqou>
the whole "it's a billing system first and foremost" attitude basically killed it
<gruetzkopf>
especially once CCS got into the game