Show Idle (>14 d.) Chans

← 2021-11-11 | 2021-11-13 →
00:03 asciilifeform verisimilitude: funnily enuff, speaking of 2014, it's what asciilifeform spent good % of it doing -- discovering how to make a reliable unwhitened rng.
00:03 asciilifeform surprisingly tricky (if no one to crib from , lol)
00:05 asciilifeform funnilyx9000, afaik to this day there's no straight patch to make trb run off fg (or any other iron trng)
00:06 asciilifeform ( various folx threatened to 'separate the wallet' , afaik no one succeeded yet )
00:07 asciilifeform billymg: pest box (esp. after we've 'rekey') defo would win from a working trng.
00:07 asciilifeform (as even stated in the spec)
00:11 asciilifeform asciilifeform's proposed rekey algo, for reference : peer A takes 512bit sA from trng, sends sha512(sA) ('key offer') to peer B. the latter does same; sB; sends sha512(sB) to A. then A sends sA to B, who verifies that it hashes to the earlier hash; if yes, sends his 'key slice' similarly to A. new mutual key is sA ^ sB ^ the key they had the conversaion with.
00:11 asciilifeform the output of the xor replaces the old key they were speaking with.
00:12 asciilifeform this way : 1) neither A nor B can impose an arbitrary (possib. weak) key 2) the entropy of the new key is the max of sA, sB, and the old key.
00:12 asciilifeform i.e. if either A or B has a proper trng, the new key is 'of trng quality'.
00:13 asciilifeform thimbronion ^
00:14 * asciilifeform still hasn't written the appropriate spec section. but the above is more or less a complete description of same.
00:15 asciilifeform ideally all peers rekey at least erry 24h
00:15 asciilifeform (or moar often when folx start piping e.g. warez over pest, and moar 'mileage wear' on keys)
00:16 asciilifeform and of course one can still send a peer new key 'by hand' / pgpgram.
00:16 asciilifeform really 'gold standard' is for key to travel rsa'd.
~ 1 hours 47 minutes ~
02:03 verisimilitude That does fix the issue I'd mentioned. If sA or sB be zero, it would no longer be disastrous.
02:04 verisimilitude At worst, the current key is still used.
02:10 signpost I separated the wallet, have a patch
02:12 signpost rather, removed, and then used pycoin to make transactions
02:12 signpost this variant of trb was part of the deedbot wallet stack
~ 1 hours 13 minutes ~
03:26 asciilifeform signpost: right, i recall. was speaking of the orig. 'cut trb in 2' concept there
03:27 asciilifeform ( not, turned out, at all simple, thing being a 1st class ball of yarn where literally erry wart interdepends on erry other )
03:30 signpost yep, didn't seem like productive use of time, when the full cut approached rewrite
03:30 asciilifeform aaha
03:30 signpost and if this, why still in c++
03:31 asciilifeform imho it is however entirely possible to transplant a sane db in place of bdb.
03:31 asciilifeform ditto the cryptonumerics.
03:31 asciilifeform aand the networking stack.
03:31 asciilifeform (and then to throw out the remaining lint...)
~ 22 minutes ~
03:54 signpost tell you hwut, this vdiff inability to handle binary files is something of a barrier to jamming existing bootloaders into a vpatch
03:54 * signpost just about ready to fart out a min-linux, but grunting through ^
04:03 asciilifeform signpost: what's wrong with e.g. echo DEADBEEF... | xxd -r -p > turd.bin ?
04:04 asciilifeform (imho this or equiv. is the sane way to represent bins in a human-readable)
04:04 asciilifeform DEADBEEF can be any length of hexolade req'd
04:05 signpost yeah, will likely end up grunting through syslinux with that.
04:05 asciilifeform if a binturd absolutely must be present -- this'd be how
04:07 signpost and actually, better this, bootsector's readable in pressed state too, and not just in patch.
04:07 asciilifeform aaha
04:08 asciilifeform i.e. in place of a binturd, you have a binturd-generator, which can be diffed. or even macroasmed into sumthing like readability, and ~then~ diffed.
04:09 asciilifeform even if all you can do to it w/out thorough reversemassage is to replace e.g. runs of 9000 0s with ZERO(9000)
04:09 asciilifeform still makes for over9000x moar 'diffability'
04:12 asciilifeform see from pov where in fact no one actually needs a bootloader, the thing folx need in reality is a proggy that will emplace a bootloader on given device
04:12 asciilifeform said proggy can very well consist of human text.
04:15 signpost was hoping to pull syslinux for this, but the thing not slim. but what one gets is boot from iso, efi, as well as bios boot.
04:15 signpost lilo, simple, but hauls in nasm
04:16 signpost whole space of bootloaders stinks of "nobody takes joy in this place"
04:21 signpost maybe for efi I just use the kernel's efi stub and efibootmgr and call it a day.
04:34 signpost << buncha these thingers just thrown in with the syslinux src in bin fmt
04:35 signpost rather than an asm txt that is compiled to c32.
~ 23 minutes ~
04:58 signpost ah wait, it was bin86, not nasm, for lilo.
04:58 signpost anyway, to bed I go.
~ 8 hours 37 minutes ~
13:36 asciilifeform << this is entirely so -- bootloader is 'unsexy', it either satisfies the bios (either legacy msdos-style, or 'modern' efi horror) and gets out of the way, or doesn't, not much room for 'high pilotage'
13:36 dulapbot Logged on 2021-11-11 23:16:08 signpost: whole space of bootloaders stinks of "nobody takes joy in this place"
13:38 * asciilifeform very partial to stuffing kernel directly in rom, but this is tricky on current-day x86 irons. worx a++ on e.g. apu.
13:41 asciilifeform interestingly, would be relatively simple (and 100% compat. with even the most 'evil' current x86 box) to keep it on 'option rom' of a pcie card. but afaik no one manufacturs such a card (ideally would have simply a hole for usb drive, to serve as the rom, and fpga to talk to the bus. update of kernel strictly by physically pulling the stick and writing to same)
13:43 asciilifeform ... then obv. would need no bootloader.
~ 45 minutes ~
14:28 punkman fun short story
14:28 asciilifeform wb punkman
14:30 asciilifeform meanwhile concluding yest's recycling operation : ty shinohai ! net yield from 'bsv' recycling into piggy == 0.09461753 . after all recycling ops, piggy nao == 34.07754558 , and this'll be in the next broadcast !
14:30 dulapbot Logged on 2021-11-11 14:37:43 asciilifeform: on shinohai's rec, trying that gox, hypothetically in ~day might see output
14:30 asciilifeform ( interestingly, the bsv eater demanded ~100~ confs )
14:31 shinohai gm asciilifeform ... yeah BSV has been 51% attacked several times in recent months, so most exchanges require many confirms.
14:32 * shinohai recommended asciilifeform perform exchange himself in case shit hit the fan and his signing key was needed ...
14:33 asciilifeform shinohai: i expect this is 1st and last time asciilifeform will need to fiddle with shitcoins, lol
14:34 shinohai But wait until the SuperExcellentRealRealBitcoin fork!
14:34 asciilifeform the lemon is 100% squeezed nao ( incidentally that same gox eats 'btg' but i was unable to find a working client for same. so ~those~ bitcents can be considered lost , likely )
14:35 asciilifeform even booted up a malware mswin box and tried their 'official' one, it wouldn't connect.
14:35 shinohai Likely 0 individuals running noads.
14:37 asciilifeform well the changely gox thing claims to eat 'btg', so presumably runs one ?
14:38 asciilifeform then again anyone can claim anyffin.
14:42 shinohai Funny enuf in most BTC forks don't even bother changing ip's in `chainparamsseeds.h` so are still attempting connection to prb nodes hardcoded in.
14:42 asciilifeform shinohai: these at one time (perhaps still, haven't looked recently) regularly popped up an' try to feed liquishit to trb noades
14:43 asciilifeform << seems to contain some kinda hieroglyphs i haven't installed in xorg, lol
14:43 dulapbot Logged on 2021-11-12 09:28:20 punkman: fun short story
~ 22 minutes ~
15:05 punkman it's littered with emojis, can ignore/strip
15:17 asciilifeform a
15:17 asciilifeform prolly wants to be read on a crapple. i'ma save for later.
~ 1 hours 20 minutes ~
16:38 asciilifeform $ticker btc usd
16:38 busybot Current BTC price in USD: $62832.74
16:38 asciilifeform !w poll
16:38 watchglass Polling 17 nodes...
16:38 watchglass : Could not connect!
16:38 watchglass : Could not connect!
16:38 watchglass : Could not connect! (Operator: jurov)
16:38 watchglass : Could not connect!
16:38 watchglass : ( Alive: (0.093s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=709399 (Operator: asciilifeform)
16:38 watchglass : Could not connect!
16:38 watchglass : Alive: (0.081s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Return Addr= Blocks=709399
16:38 watchglass : ( Alive: (0.082s) V=70001 (/ Jumpers=0x1 (TRB-Compat.) Blocks=709399
16:38 watchglass : ( Alive: (0.107s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Return Addr= Blocks=709399
16:38 watchglass : Alive: (0.144s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=709399 (Operator: asciilifeform)
16:38 watchglass : ( Alive: (0.179s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=709399
16:38 watchglass : Alive: (0.088s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Return Addr= Blocks=709399 (Operator: whaack)
16:38 watchglass : Alive: (0.225s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=709399
16:38 watchglass : ( Alive: (0.253s) V=88888 (/ Jumpers=0x1 (TRB-Compat.) Blocks=709399
16:38 watchglass : Alive: (0.245s) V=70001 (/ Jumpers=0x1 (TRB-Compat.) Blocks=709399
16:38 watchglass : ( Alive: (0.595s) V=99999 (/ Jumpers=0x1 (TRB-Compat.) Blocks=709399
16:39 watchglass : Busy? (No answer in 100 sec.)
16:44 * cgra wears scoopbot mask and: "New post on cgra's: TRB Defect Exhibition - Two DoS Classics"
16:44 asciilifeform ohey
16:47 * asciilifeform commented, tho atm only re sect.2
16:54 asciilifeform cgra: in practice, is ~impossible to use 'cooked' faux-old blox to perma-wedge a node -- the victim would have to be connected to the perp and no one else. but they do waste substantial cpu time.
17:06 cgra asciilifeform: why has to be connected to perp only?
17:07 asciilifeform cgra: if yer connected to a genuine noad, eventually will reorg to proper chain
17:07 asciilifeform then again nobody knows how long a reorg is actually possib. in practice.
17:08 * asciilifeform not yet observed in the wild a noad wedged by being fed a constructed >pow fauxchain. dun mean it's impossible.
17:08 cgra ah you mean can't fool into inhabiting an altchain, as in to believe were the main chain
17:08 asciilifeform cgra: can, tho, as i understand, waste astonishing amt of cpu time & disk tho
17:09 asciilifeform (esp. given as current trb stores blox that got reorg'd away)
17:09 cgra just meant as another fill ram/disk DoS
17:09 asciilifeform aha
17:09 asciilifeform imho 'cement' is an adequate pill.
17:09 asciilifeform still gotta be implemented tho
17:09 cgra asciilifeform: yeah, should be easy to implement
17:10 cgra how about user knob of 'how much work or height to past at most, considered valid'?
17:11 asciilifeform hm?
17:12 cgra asciilifeform: iirc you intended similar for nqb, the line between forever frozen blocks and possibly-reorging blocks
17:12 asciilifeform aa
17:12 asciilifeform i.e. permit operator to request a noad to consider 0...n blox as already cemented
17:12 asciilifeform potentially risky, he has nfi what they are necessarily
17:13 cgra asciilifeform: i suppose it's something the operator needs to keep an eye on, perhaps a runtime knob?
17:13 asciilifeform cgra: indeed asciilifeform proposed a max realistic reorg for nqb, but strictly for allowing readonly o(1) blox db
17:15 asciilifeform trb already has a max reorg length, we simply dun know what it is, lol
17:16 cgra interesting point :)
~ 18 minutes ~
17:35 cgra << to be exact, permit the operator to tell node how much recent work reorgable. this seems to me like it's a thing 'operator must operate' like periodic adding to the cement would be. if your comment still applies, can you elaborate?
17:35 dulapbot Logged on 2021-11-12 12:12:24 asciilifeform: i.e. permit operator to request a noad to consider 0...n blox as already cemented
17:35 dulapbot Logged on 2021-11-12 12:12:37 asciilifeform: potentially risky, he has nfi what they are necessarily
17:36 asciilifeform cgra: imho there defo oughta be such a knob. simply noting that it aint a 100% substitute for 'cement' where can specify concrete known hashes.
17:36 asciilifeform (incl. to massively speed up rejection of bogoblox during init sync)
17:37 cgra right
17:37 asciilifeform a cemented noad oughta sync at ~linerate -- it knows exactly what to ask for (and what to throw out w/out even reading)
17:42 cgra asciilifeform: is 'line rate' assuming other block processing bottlenecks fixed, or hash optimism?
17:45 asciilifeform cgra: main bottleneck historically is bogoblox per se
17:46 asciilifeform (they aint even always bogus per se, simply 'not the next one')
17:46 asciilifeform btw imho a noad oughta still verify a block normally even if its hash was found in 'cement'
17:46 cgra asciilifeform: swallowing a block even on an intelist box, seconds/piece
17:47 asciilifeform cgra: some up to minute+, on e.g. asciilifeform's apu1 noad
17:48 asciilifeform replacing the db with proper o(1) nqb+cryostat would fix this, bdb is monstrously slow
17:48 asciilifeform main bottleneck historically however is the low snr, so to speak, in incoming blox during init sync
17:48 asciilifeform (i.e. not only is db slow, but most of the churn constitutes wasted cycles on unwantex blocks)
17:48 asciilifeform *unwanted
17:49 cgra yeah
~ 2 hours 52 minutes ~
20:42 asciilifeform hm thimbronion i'm on pestnet (testing a rubbish-sending variant of your current vdiff) and send a helloworld, but not showing up in pestlog
20:44 asciilifeform the draft patch, for reference.
20:47 asciilifeform oh lol it lost my keys didntit
20:47 asciilifeform disregard patch, it is braindamaged, will need to redo
20:50 thimbronion asciilifeform: ack
20:50 thimbronion good to know you are working on such
20:54 asciilifeform thimbronion: how on earth do i get current own nickname from inside 'server' ?
20:59 asciilifeform the oopism is ultra-hobbling
21:03 asciilifeform why aint there a send_broadcast which takes arbitrary payload ?
21:03 * asciilifeform prolly oughta leave this alone until thimbronion gets a chance to sweep it a little
21:04 asciilifeform atm mega-spaghetti (with all respect to author, who baked it in record time)
21:04 * bonechewer is unqualified to write a debunking of the oopism present in nearly every introductory programming text, but would love to read a good one if it exists
21:05 asciilifeform bonechewer: the most persuasive 'debunking' is an encounter with typical oop proggy, lol
21:06 bonechewer yeah but hard to explain to the rank beginners that I work with just WHY they should ignore that chapter with the usual useless contrived example of "car" inheriting from "vehicle", etc, etc.
21:06 bonechewer I've found oopism useful for GUI stuff, rarely if ever elsewhere
21:07 asciilifeform bonechewer: oopism as typically encountered was pushed by salt mine types, and for logical reasons -- it ties the hands of the mediocrities they employ to try to limit the damage that can be done by a particular indian shudra
21:07 asciilifeform (by isolating subsystems)
21:08 asciilifeform rather like the plastic scissors given to american schoolchildren
21:08 asciilifeform in very few places is oopism a net win from any other pov
21:09 bonechewer that sounds about right, but doesn't explain why every damned learn-to-coad book feels obligated to posit oopism as essential core knowledge
21:10 asciilifeform bonechewer: for simple reason. the books are largely written for aspiring 'indians' .
21:11 * thimbronion has spent many an hour staring into the server.rb/client.rb spaghetti bowl
21:12 asciilifeform thimbronion: if you're able to make a working variant of my patch, would be much appreciated. presently asciilifeform is at a loss re how, the req'd knobs aint in 1 obv. place.
21:13 asciilifeform thimbronion: idea is to make use of the pingpong timer as a kludge for getting a minimal rubbish flow rate to keep ephemeral port from evaporating.
21:13 thimbronion I will check out the patch. To get a nickname, you need to have access to a particular client instance.
21:13 asciilifeform (a user-controllable timer would be Right Thing, tho)
21:13 asciilifeform ugh ok
21:13 asciilifeform that's what i meant re oopism
21:13 bonechewer separately: wrt BTC-accepting colo, I have no personal experience with but some internet acquaintances recommend it as cheap and flexible.
21:13 dulapbot Logged on 2021-11-11 16:23:19 asciilifeform: whaack: i defo agree re nodes. plox to write in with recs if you know of reasonable quality btc-eating hosts (other than asciilifeform's, lol, cage)
21:14 asciilifeform bonechewer: lol their www times out
21:14 asciilifeform not a good sign
21:15 PeterL asciilifeform: for something like this, how would you structure the program without using oopisms?
21:15 asciilifeform thimbronion: imho there's no good reason to have a pest instance support >1 irc client. (i can guess why did this, but imho guest users are best implemented using an external proggy)
21:16 bonechewer hmm, works for me... though hopefully they have enough sense to host their www outside their own data center
21:16 thimbronion asciilifeform: noted. I didn't add that capability in myself, just came with the irc server code I repped off, which naturally would support multiple clients.
21:17 asciilifeform bonechewer: it dun load here
21:17 asciilifeform at all
21:17 bonechewer sigh
21:17 asciilifeform $ curl -v
21:17 asciilifeform * Trying
21:17 asciilifeform * Connected to ( port 443 (#0)
21:17 asciilifeform * ALPN, offering h2
21:17 asciilifeform * ALPN, offering http/1.1
21:17 asciilifeform * successfully set certificate verify locations:
21:17 asciilifeform * CAfile: /etc/ssl/certs/ca-certificates.crt
21:17 asciilifeform * CApath: /etc/ssl/certs
21:17 asciilifeform * TLSv1.3 (OUT), TLS handshake, Client hello (1):
21:17 asciilifeform hangs.
21:17 asciilifeform thimbronion: makes sense
21:18 bonechewer well, that is indeed probably a bad sign. "works for me" is probably not the kind of routing one wants.
21:18 asciilifeform bonechewer: prolly 1 of those morons who blocks half the planet because got ddosed at some pt
21:23 * asciilifeform must bbl
~ 17 minutes ~
21:41 thimbronion asciilifeform: I see what you're doing with the patch. It conflicts with some changes that are underway. I will wrap those up and get a version of your patch in next.
21:48 billymg << fwiw at the moment my pest instance has two clients, myself and the bot
21:48 dulapbot Logged on 2021-11-12 16:15:48 asciilifeform: thimbronion: imho there's no good reason to have a pest instance support >1 irc client. (i can guess why did this, but imho guest users are best implemented using an external proggy)
21:53 shinohai I'm using to separate instances for my bot and self.
21:53 shinohai s/to/two/
22:01 billymg shinohai: same box?
22:01 billymg or different boxes also
22:04 shinohai billymg: different boxes.
22:06 billymg shinohai: gotcah
22:06 * billymg needs more boxes
~ 1 hours 3 minutes ~
23:10 asciilifeform << thx thimbronion
23:10 dulapbot Logged on 2021-11-12 16:41:10 thimbronion: asciilifeform: I see what you're doing with the patch. It conflicts with some changes that are underway. I will wrap those up and get a version of your patch in next.
← 2021-11-11 | 2021-11-13 →