Show Idle (>14 d.) Chans

← 2020-08-19 | 2020-08-21 →
mats: great new pic on phfs log
trinque: mats: link?
asciilifeform: mats: lulzy
asciilifeform: !w poll
watchglass: Polling 12 nodes...
watchglass: : Alive: (0.082s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=644567
watchglass: : Alive: (0.085s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=644567 (Operator: asciilifeform)
watchglass: : ( Alive: (0.099s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=644567 (Operator: asciilifeform)
watchglass: : ( Alive: (0.082s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=644567
watchglass: : ( Alive: (0.143s) V=70001 (/ Jumpers=0x1 (TRB-Compat.) Blocks=644567
watchglass: : Alive: (0.166s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=644567
watchglass: : Alive: (0.212s) V=70001 (/ Jumpers=0x1 (TRB-Compat.) Blocks=644567
watchglass: : Alive: (0.229s) V=70001 (/ Jumpers=0x1 (TRB-Compat.) Blocks=644567
watchglass: : Alive: (0.377s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=644563
watchglass: : ( Alive: (0.396s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=644567
watchglass: : ( Alive: (0.587s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=644567
watchglass: : Busy? (No answer in 20 sec.) (Operator: jurov)
Aerthean: Hi Asciilifeform, I've been a reader of the logs and your website for several years.
Aerthean: I grabbed a copy of interlisp based on what Amberglint mentioned yesterday, and I found the lfg software for it (dunno if this is the OS soft you're looking for) in an archive of Xerox PARC's ftp server from 2014.
Aerthean: There's a decent chance you already came across this, but figured I would mention it on the chance you hadn't.
Aerthean: Thanks for the work you've done on FFA and your other writings.
asciilifeform: neato. and yer welcome.
asciilifeform will look when there's time
asciilifeform: welcome to #a btw, Aerthean. plz consider sitting long enuff to have conversation...
asciilifeform: << funnily enuff, seems that i had it on my hdd, dated july 2010...
snsabot: Logged on 2020-08-20 13:26:11 Aerthean: For the lfg.sysout in particular it's located at
asciilifeform: ... bitwise identical to linked item.
asciilifeform: must've dug it up when last played w/ 'medley'.
asciilifeform: !!key Aerthean
deedbot: Not registered.
asciilifeform: Aerthean: consider getting in WoT.
Aerthean: Sorry, I wasn't sure about the etiquette of leaving a connection open from somewhere or just following up on responses that appeared in the logs
Aerthean: Regarding registering a key with the WoT, does a process exist for updating said key post registration (aside from asking trinque? nicely)
asciilifeform: Aerthean: it's done by asking trinque nicely (and obv. by signing said req w/ the key being replaced, as well as the successor key)
asciilifeform: Aerthean: there is deliberately no automatic mechanism
asciilifeform: Aerthean: re the other q -- it makes 0 diff to me whether folx sit logged in 24/7, or only come to answer logs. the 1 historic annoyance is noobs who think they can 'get sense of what chan is about' by idling for a day. but in your case this obv. not the case.
Aerthean: Thanks, I thought that was the case, but better to verify. I could generate a key on my current computer, but I may be better off delaying until I bring something up more stripped down. Even though I've following the logs, and know how to act better security-wise, still often act lazy.
asciilifeform: Aerthean: the traditional state of the art is to uncrate an ancient laptop and do all pgp ops (incl. generation) there.
Aerthean: I bought one of the PCEngines APU1 for that purpose
asciilifeform: the key never leaves the box. i/o via serial port, or, if lazy, removable drives.
asciilifeform: apu1 will work, tho it makes entering passphrase painful (no vga)
Aerthean: I've also got a few old laptops floating around, I think at least one of them pre-dates the Management Engine
Aerthean: Either way, I'll get something set up, and get a key registered.
asciilifeform: btw asciilifeform's x64 gentoo worx on apu1 w/ only 1 modification ( need s0:12345:respawn:/sbin/agetty -L 115200 ttyS0 vt100 in /etc/inittab )
Aerthean: Thanks, it's worth setting that up on the apu1 whether I use it for keys or not
asciilifeform: Aerthean: a++. when you register, can get ratings, and then do biz , meaningfully sign code , and many other things.
asciilifeform: Aerthean: if you use this gentoo, you will probably also want to use my packages repo . in /etc/portage/make.conf , GENTOO_MIRRORS="" .
asciilifeform: ( this is supposing you want to use 'portage' at all )
asciilifeform: it aint in the installer by default, i baked the latter before had the mirror.
Aerthean: I recall you had a strong warning regarding emerging and syncing. Given the rather large changes gcc that you've mentioned ( removing of static linking, etc )
asciilifeform: Aerthean: actually this gentoo still based on traditional glibc. static linking in it still has to be done by building a separate gcc + toolchain, or using ave1's gnat , or similar.
asciilifeform: Aerthean: however my gentoo is built on gcc 4.9, and carefully cleansed of systemdism. as well as being 'update-proof'.
asciilifeform: Aerthean: trinque is working on a purely musl-based linux & toolchain.
asciilifeform: ( incidentally -- trinque , shinohai , et al -- ave1's www seems to be dead nao.. )
asciilifeform: !q seen-anywhere ave1
snsabot: ave1 last seen in #ossasepia on 2020-04-06 04:23:33: So could you try to build with single worker?
Aerthean: Ah, yeah I followed along with some of the musl work, I put some effort into building a musl-based FreeRTOS system for a Xilinx Zynq
asciilifeform: Aerthean: building a musl-based gcc toolchain is not so hard; the harder thing is getting 100% of proggies to actually build in it (for some, trivial, for others -- e.g. emacs -- yet-unsolved afaik , some of'em use glibcisms specifically )
Aerthean: Eventually I gave up as I had other work to do, and didn't have the motivation to find and tear out all the Xilinx'isms that prevented it from building
asciilifeform: xilinx is epic ball of yarn, entirely comparable to microshit's products
Aerthean: Yeah, I discovered that, just a bunch of miscellaneous differences
Aerthean: Yes, I use their software for work
Aerthean: I think they have a separate copy of their Place and Route software (or however one would describe it), for every single FPGA instance
asciilifeform: i also used , for a hardware product , as well as earlier for various experiments.
asciilifeform: Aerthean: i personally have at least 3 copies of their toolchain, and pretty sure that support for all the chips i ever used aint even in their current release..
Aerthean: With Vivado they pulled out support for most of their older chips
Aerthean: Spartan 6's in particular
asciilifeform: the 9572 i think vanished almost decade ago from the official package
snsabot: Logged on 2020-08-20 17:59:10 asciilifeform: ( incidentally -- trinque , shinohai , et al -- ave1's www seems to be dead nao.. )
asciilifeform: ty shinohai
Aerthean: For an updated FG, I do a fair amount of work using LYSO + SiPMs for various things
Aerthean: You've mentioned using LYSO in the past in the logs, and I've been curious (in general) what approach you planned on taking
asciilifeform: Aerthean: built prototype where simply debias a 250MHz counter b/w events. but not satisfied with my amplifier (requires rather costly PIN diode, and very good shielding) , project currently backburnered
asciilifeform: Aerthean: what also want, ideally, is a 100% analogue filter to count ~only~ the L decay events (i.e. the exact opposite of what folx who use LYSO scintillators normally do.)
Aerthean: Ah, yeah the decay events tend to get in the way, especially if you're using Na22
asciilifeform: Aerthean: idea is, the rng should not depend on external events at all.
Aerthean: So you'd have it sealed and depend solely on the L decay events?
asciilifeform: correct.
asciilifeform: Aerthean: currently this item not a priority for asciilifeform , as his previous attempt to market a much cheaper, wholly electronic rng was not a success.
asciilifeform: there simply aint many folx who want these badly enuff to pay what they'd cost.
Aerthean: Yeah, that's true
asciilifeform: this is actually a problem that plagues virtually any attempt at 'sane irons' projects.
Aerthean: Cost-wise how much were you looking for the electronics in the system?
asciilifeform: esp. if yer contemplating ic fabbing.
Aerthean: Not a cheap business, and things always go wrong
asciilifeform: Aerthean: i got a chinese firm to grow the LYSOs , <50$/ea in reasonably small qty. electronics, i expect, would cost similar. but shield is a mechanical object, and costs substantial $ to fabricate. likewise (and this was problem w/ FG as well) testing, it ultimately has to happen on my desk and the time aint free.
asciilifeform: the current design also relies on a PIN diode, and the cheapest uv-optimized one i was able to make use of reliably is ~150$ in small qty.
asciilifeform: generic PIN diodes have very poor (borderline unusable) SNR , i found.
asciilifeform: at least for this application.
Aerthean: I've found PINs good for non-fast pulse related tasks
asciilifeform: for comparison, unit cost of FG , in materiel, was <50 $ . and still not 'bestseller'.
asciilifeform: Aerthean: aah, but the lower the deadtime, the better bitrate of rng.
asciilifeform: ( and asciilifeform realizes that 100% of folx who traditionally care about fast pulse, use PMTs. but i rejected'em for same reason as rejected geigers -- they require HV, which requires oscillators; and they ~age~ detectably, which is unacceptable in rng )
Aerthean: True. I have some PINs at the lab and some LYSO tiles I'll try to take a look and see what results I get
Aerthean: Have you considered using SiPMs?
Aerthean: Does require bias
asciilifeform: Aerthean: the 150 $ item is a SiPM ( i.e. micro array of PINs )
Aerthean: Depending on SiPM 40-70ish volts
asciilifeform: and yes req's bias. i generated the bias via a cheat, that some might find interesting -- a set of sealed photoisolators, powered in parallel but outputs in series
asciilifeform: extremely inefficient but afaik the ~only~ way, other than battery power, to get +45v w/out oscillator anywhere
Aerthean: What did you use for the amplifier circuitry?
asciilifeform: i'd dearly luvv to come up w/ a way to avoid using SiPM...
asciilifeform: Aerthean: differential amp
asciilifeform: Aerthean: if/when actually contemplate selling these, will publish the schems, just like published FG's
Aerthean: You can get fairly good results from using a Sziklai Darlington
Aerthean: MMIC amps work as well, but suck a lot of power
asciilifeform: i suspect that all possible ways to make the amp will share the annoyance of requiring fiddling w/ trimpots and scope during assembly
asciilifeform: and a large and very tight shield
asciilifeform: hence would have to sell these for equiv. of coupla $k to come out in the black.
Aerthean: I've run boards with SiPM + LYSO in open air, no shielding with good results
Aerthean: Just light covering
Aerthean: no metal shield
asciilifeform: Aerthean: right, but presumably you weren't trying to make trng.
Aerthean: Also true
asciilifeform: the req's are completely different from e.g. 'get rough count of atmospheric gamma'
Aerthean: I'm assuming you're trying to keep it from leaking RF everywhere
asciilifeform: a trng that even in theory can be influenced by nearby rf, is a rubbish
asciilifeform: leaking yes, but primarily ~from receiving~
asciilifeform: most naive attempts at electronic rng, for instance, are really picking up local radio
asciilifeform: in FG, the shields were actually substantial part of the unit cost.
asciilifeform: ( and substantial effort went into shortening traces, and getting something like a solid ground plane while keeping the pcb 2layer for cost )
Aerthean: Hm, it's odd, because I've used SiPM + LYSO and while you have some noise, it's usually worst case the size of the SiPM single-photon or dark pulses
Aerthean: LYSO pulses were much bigger than that
asciilifeform: Aerthean: put yer amp next to e.g. keyed ham radio , and see what happens.
asciilifeform: BingoBoingo: maybe i'm thick, but what did he mean in 'click on the tag' ?
BingoBoingo: asciilifeform: Up top he has a link to the category he uses for book reviews he has done
Aerthean: Ah! There's the misunderstanding, you want robust against RF attack
asciilifeform: Aerthean: correct.
asciilifeform: Aerthean: including 'attack' where attacker simply is aware of nearby strong signals.
asciilifeform: e.g. if it's picking up 'voice of america', it aint a trng.
Aerthean: Well, I can't say robust against say ham radio next to amp, but I can say doesn't pick up FM stations
asciilifeform: BingoBoingo: re lafond -- i actually had bought a pile of his boox, coupla yrs back when you linked him ~erryday. they, i confess, were a snore, largely endless cut&pasta from his www. extremely repetitive.
asciilifeform: Aerthean: whole point of using scintillator, is to remove, as much as physically possible, any external influence on the rng output, either manmade or natural. external meaning from anywhere outside the box.
asciilifeform: this includes power supply. (in purely electronic rng, yer stuck w/ battery power if you want troo isolation in the above sense.)
asciilifeform: << shinohai , sadly this snapshot includes ~none of the useful zpf material. or even the last published ver. of his gnat.
snsabot: Logged on 2020-08-20 18:06:43 shinohai: <<< Archive link to his blog for teh logz
asciilifeform: shinohai: for the former, phf's www has the patches but not of course the discussion.
shinohai: I should have posted for all the url's that have snapshots
shinohai: But anyone with grey matter can do.
Aerthean: In the end, best thing to do is run some tests and post data for others to look at
asciilifeform: Aerthean: not to say that testing is useless (see the various folx who tested FG, linked from the preserved www ) but some problems can only be dealt with properly ~categorically~, i.e. by making their occurrence physically impossible by design.
asciilifeform: Aerthean: for instance, i avoided a whole world of hurt in orig. FG design by eschewing traditional avalanche diode , which in turn needs HV , and clean source of it
asciilifeform: thing could run off 2 AA cells in fact.
Aerthean: I remember reading the paper you posted about an RNG that used a Zener I think.
asciilifeform: prolly the oldest design known
Aerthean: Except they messed up their design by whitening the output
asciilifeform: right, afaik erry trng ever marketed, other than mine, committed this sin
asciilifeform: zener, even if one powers it as i did (expensively) with the SiPM, also suffers from age effects.
asciilifeform: for all the surface simplicity of trng, it is surprisingly hard to get 100% right, esp. if you are unwilling to lie to yourself & others via 'whitenings'.
Aerthean: I've found that a truism about many things.
asciilifeform: indeed applies to many things.
asciilifeform: it is also the case that rng as commercial product is a very questionable biz proposition. it takes quite a bit of 'adulthood' to even get to a place where you actually benefit from a 1000 $ rng. for instance, microshit victims dun really win anyffin from using whatever external rng.
asciilifeform: it's a little like marketing, say, a home autoclave. most folx will still not be qualified to perform surgery even if the tools were priced to be available to them.
Aerthean: Yes, just because one has a tool doesn't mean one can use it correctly
Aerthean: Man with hammer syndrome
asciilifeform: Aerthean: right. if i ever do sell the lyso rng, it'll be part of e.g. self-contained OTPtron, or similar.
Aerthean: Thanks for indulging my curiousity.
asciilifeform: Aerthean: no prob, and dun hesitate to ask in future.
asciilifeform: fwiw i made the design of the earlier attempt pedantically, agonizingly public, there are literally kilometers of log discussing it
asciilifeform: and 100% of sources/schems on www still, long after shop closed.
Aerthean: I downloaded them in the past and read through them. Thought about baking a copy
asciilifeform: Aerthean: it aint terribly hard, esp. if you dun need it to be compact so to fit in servers
asciilifeform: original prototype was breadboarded (the 9572 was on a 'sample pack'-type DIP adapter)
Aerthean: The parts also don't cost much
asciilifeform: the analogue piece does have to be shielded tho
asciilifeform: indeed they do not cost much.
asciilifeform: i published the design in 2016, but no one has written in with 'i replicated..'.
Aerthean: As mentioned above, most don't know of its existence. Then of those who do know, most probably don't care enough to build / buy.
asciilifeform: Aerthean: near as i can tell, most computer users who even are aware of concept of rng, are satisfied with 'homeopathic' ones.
asciilifeform: hence 'lemon market'.
asciilifeform: there are some euro firms who offer 20,000+ $ 'golden toilet' rng. i investigated, all of'em whiten.
asciilifeform: i suspect they are simply tentacles of nsa-controlled 'crypto ag' shell company.
Aerthean: Given the amount of money sloshing around these days it's hard to tell
Aerthean: Nth front corporation of some US / other govt. organization
asciilifeform: Aerthean: related reading. it is ~very~ difficult to profit by 'actually Do Right Thing' in engineering, vs. 'convincingly pretend to Do Right Thing' , 'i can't believe it's not butter!' atrocities.
Aerthean: Ah hah, very good piece
Aerthean: I think one can only do the right thing if one has a robust meat + extended wot
Aerthean: For the record, I don't. So I continue to work my simple technician job.
asciilifeform: in principle anyone can do. what can't do, is run a biz on such product, an' stay in the black.
asciilifeform: Aerthean: i also day job.
asciilifeform: Aerthean: exactly 0 of my attempts at 'Right Thing' ever so much as threatened to make it unnecessary. and i suspect is true for most folx who attempt.
Aerthean: I suspect if it were becoming a success, then something would happen
asciilifeform: Aerthean: very rarely does anyone have to manually gas people for 'right thing', normally market puts'em outta biz 'naturally'.
asciilifeform: as discussed in the linked orlov piece.
asciilifeform: this is why you can't buy even a toaster that lives for > coupla yrs.
Aerthean: True, even natural events can bankrupt you
Aerthean: You depend on a particular IC and the vendor stops building it
Aerthean: See: I'm sorry for your loss.
asciilifeform: Aerthean: this is part of why i used cpld, rather than microcontroller
asciilifeform: theoretically, 0 vendor lock
asciilifeform: could even have the circuit fabbed. if there were actually serious demand for it.
asciilifeform: see the src, i specifically did not use ~any~ coad from xilinx's libs.
asciilifeform: my original intent in fact was to bake it as a board fulla 74xxx standard discretes. but this wouldn't fit in a 1u.
Aerthean: Very easy to examine then
asciilifeform: the more general idea was discussed in the logs many times. it is why you can't buy a sane, open cpu. there is simply no way to pay for what the fabrication would cost. esp. if you refuse to go along with the vendor's demands to use their NDA'd cell libs & mswin shitwares.
asciilifeform actually went an' investigated a number of firms that offer 'small-run' ic fab in various processes. they all have these in common.
asciilifeform: ftr what asciilifeform was interested in fabbing, was simply a homogeneous-fabric fpga, scaled-up version of e.g. classic ice40 .
Aerthean: I only have small knowledge of the fabrication field, mostly based on researching what you + others wrote about
asciilifeform: there is presently no 100%-documented homogeneous fpga on market. and, for fundamental reasons, aint likely to be, 100% of fpga vendors live from the 'rent out the right to use our built-in NIC' and similar scamola
asciilifeform: Aerthean: you'll find various figures for cost, but keep in mind that almost all ic fab projects involve multiple attempts.
asciilifeform: and full-time designers.
Aerthean: I've done some work with a company that got some non-silicon analog devices fabricated and I helped with the wafer dicing and testing. This particular product never really worked out
asciilifeform: the small-run folx also typically do not include ~packaging~ in their cost quote. (usually people who shell out for ic proto runs have the necessary gear in-house.)
asciilifeform: they also offer 0 guarantees of yield.
Aerthean: Yes, in our case the yield turned out ok, but the design had other problems.
asciilifeform: right, hence 'take the quote and multiply it 10x or however many revisions it could take'.
Aerthean: The company also had their own small fab where they could modify / bake things in very tiny quantities
Aerthean: The people we worked with
asciilifeform: unsurprising
Aerthean: Probably the only way to do anything cost-effectively
asciilifeform: presently i'm not aware of even a single instance of a fab job financed and successfully carried out by a 'human' (vs at least medium-sized organization).
asciilifeform: dun mean that it's impossible -- certainly possible if yer sitting on a coupla $mil. but asciilifeform aint.
Aerthean: and when you get a couple of $mil, suddenly other things become important
asciilifeform: Aerthean: the folx who are, typically are concerned with ROI.
Aerthean: Yes, that or they'd rather spend the money on drugs, whores, houses, boats, cars, etc
asciilifeform: correct.
asciilifeform: that's where asciilifeform's old coauthor eventually went.
asciilifeform: i can't really blame folx for not wanting to throw $mil down black hole.
Aerthean: I never showed up in trilema as MP started getting very ban happy
asciilifeform: Aerthean: imho was quite palpably 'fuhrerbunker 1945' for most of 2019.
asciilifeform: ( and 2020 -- fin. )
Aerthean: I found his writing hit or miss, more miss in the future.
asciilifeform: if seen film 'die untergang' -- will know, good illustration.
asciilifeform: 'germany is not worthy' etc.
Aerthean: Yes, I have seen that film
BingoBoingo: << Autoclaves do find their ways into homes of fungiculture enthusiasts.
snsabot: Logged on 2020-08-20 19:02:27 asciilifeform: it's a little like marketing, say, a home autoclave. most folx will still not be qualified to perform surgery even if the tools were priced to be available to them.
asciilifeform: BingoBoingo: i'm aware.
asciilifeform: BingoBoingo: possibly better example would be the idjits who bought up respirators during 'plague panic'
BingoBoingo: Bipeds will appreciate civilized techs to the extent it allows them to dope
asciilifeform: BingoBoingo: afaik the magic mushroom folx simply use pressure cookers tho.
BingoBoingo: Ah, that yes. Especially considering those respirators drove a lot of the early mortality during the panic. (mechanical ventilation + inflamed lungs = traumatized and inflamed lungs)
asciilifeform: aha, lungs popped like balloons.
BingoBoingo: asciilifeform: Some do, but dope people do their own sort of inner circle stotting. Same thing drives increasingly elaborate water pipes.
asciilifeform: BingoBoingo: i'm only surface-aware of what goes on in that subculture, but can picture.
feedbot: << The Montevideo Standard -- USSA: Judge Gives Offender Probation And "Anger Management" In County Fair Killing
BingoBoingo has been proceeding these past few months as though tourism and entertainment, however anti-socially reconfigured, will return in a way.
Aerthean: BingoBoingo: Aside from water pipes the last time I had contact with the dope subculture vaping? had become popular. Filling a large plastic balloon with smoke and breathing it in.
asciilifeform: BingoBoingo: just how 'went away' is it in yer orcistan ?
BingoBoingo: asciilifeform: Border largely closed. Brothels reluctant to formally reopen. Theaters and museums adopted a bunch of weird.
feedbot: << The Montevideo Standard -- 90% Of Workers With Unemployment Insurance Expiring In July Back To Work
BingoBoingo: Aerthean: Vaping takes rather heavily tariffed tech here. Papers are still the locally dominant smoking tech.
Aerthean: BingoBoingo: Thanks, I'm surprised they would have such high tariffs on it.
BingoBoingo: Aerthean: 400 USD balloon filler turns into 900 USD balloon filler
BingoBoingo: Aerthean: General thing they do to imports here.
BingoBoingo: Computers, pnoHes similarly marked up in the general case.
Aerthean: BingoBoingo: That's a high markup! I was tangentially aware of the tariffs from the Pizarro venture, but in my mind I consider balloon filler fairly low tech
BingoBoingo: Other devices far less expensive, but the balloon filling one still fairly fringe first world thing. Gets tarriff markup and "This is going to take up shop space for a while" markup
Aerthean: Ah, every middle-man wants his cut. I recall first seeing it in the states back in 2010-2012ish.
mats: i have a few fg i wouldn't mind selling for 100usd of btc
shinohai: mats: Will DM you back about that in a few days.
trinque: << correct, but I'm pretty nice about it. just don't abuse it.
snsabot: Logged on 2020-08-20 17:25:56 Aerthean: Regarding registering a key with the WoT, does a process exist for updating said key post registration (aside from asking trinque? nicely)
trinque: << meh, I'm done mourning these. waste abounds.
snsabot: Logged on 2020-08-20 17:59:10 asciilifeform: ( incidentally -- trinque , shinohai , et al -- ave1's www seems to be dead nao.. )
Aerthean: trinque: Thanks, I will attempt to not have to change my key. Currently going through ascii's dulap instructions, once I get things working I'll generate a key and register it.
← 2020-08-19 | 2020-08-21 →