log☇︎
▁▁▁▁⏐︎▁▁▁ 3330
asciilifeform: http://btcbase.org/log/2018-04-22#1804633 << sadly asciilifeform is not so familiar with the existing proggy, to answer this properly ☝︎
a111: Logged on 2018-04-22 21:29 mircea_popescu: asciilifeform, in practical terms, how do you want s.mg to use fgs until rewritten ?
mircea_popescu: yeah. anyway, i'm thinking the best approach would actually be a kernel patch to destroy the extant random/urandom bs and replace it with fg
mircea_popescu: ideally simply have it exposed as /dev/random
asciilifeform: right
asciilifeform: would have to properly handle all types of attachment for the fg tho ( usb serial, genuine serial, etc )
asciilifeform: and so would be sensitive to device init order, which imho is icky
asciilifeform: asciilifeform often wonders if we're doomed to make a pci fg
asciilifeform: ( why not made originally ? pci adds 100x the complexity; ugly; expensive )
asciilifeform: the pc arch just plain old suxxxx. every method of dealing with this, has own tradeoff.
asciilifeform will admit, to having sat down to write a kernel mod for fg no fewer than 3 times, and barfed each time
asciilifeform: https://archive.li/xhVBU << in related vintage entomology .
asciilifeform: ( for the sake of record : a non-retarded arch would do random like-so : ~as a cpu instruction~ that reads from N ~asynchronous~ ports , where N is odd, debias, and xor )
asciilifeform: speaking of fg, ave1 really oughta have gotten his fg crate by nao
asciilifeform: ave1 didja get it ? ☟︎
asciilifeform: the 2nd attempt at the crate went out on the 11th
asciilifeform: https://elixir.bootlin.com/linux/latest/source/drivers/char/random.c << the 'traditional' /dev/random. ( get yer barf bag ready! )
asciilifeform: pointerolade, spinlock, sha1, you name it , it's got it...
asciilifeform: and 2200+ln
asciilifeform: ( not including #include'd rubbish, of which there is a mountain, as is typical )
mircea_popescu: why ? if it doesn't get juice it just blocks
mircea_popescu: you're more than welcome to fix it.
asciilifeform: which why?
mircea_popescu: why handle anything.
asciilifeform: because afaik shitkernel offers no deterministic way to fix
asciilifeform: ( if i'm wrong about this, folx, plox to write in )
mircea_popescu: doesn't have to. /dev/random is blocking. all it needs to do is block if it has no fg.
asciilifeform: right but how does it know where is an fg.
mircea_popescu: it does not care.
asciilifeform: from where does it read ?
mircea_popescu: from wherever we make the standard. usb i guess ?
asciilifeform: right but where.
mircea_popescu: what do you mean, where.
asciilifeform: could implement module the obvious way : takes N args, each of which must at load time be a valid /dev/ttyxxxx
mircea_popescu: yes.
asciilifeform: but this is incompatible with, say, a trinque-style kernel, that dun support loadable modules at all
mircea_popescu: well if the kernel can't be patched then a patch won't help.
asciilifeform: ( i have the rockchips currently with these )
asciilifeform: mircea_popescu: the current set of kernel modules, largely work in either 'module' or 'built-in' mode, depending on how compiled
mircea_popescu: but the idea isn't for ~our~ kernels. the idea is to have an infection vector, that permafucks a linus-tso kernel into no longer working like a piece orf shit.
asciilifeform: the scheme described above, would work only as a loadable mod.
asciilifeform: plox to elaborate?
mircea_popescu: i don't believe in this philosophy of "universal support". "break everything until it fully conforms to tmsr expectation" is entirely sufficient.
asciilifeform: recall, linus's kernel has no notion of random, it implements '/dev/random' as yet another sort of faux device, via module
asciilifeform: mircea_popescu: this sounds great and wholesome but still gotta decide exactly how to break and where.
mircea_popescu: indeed. which is why i was encouraging mod6 to look into it.
mircea_popescu: but let's delve into detail here :
asciilifeform: i can describe some known dead ends. for instance, you definitely do not want to marry the thing to the pl2303. because 1) you have no way of knowing that every pl2303 on usb is an fg 2) not every fg user got a pl2303 , they get whatever cheapest chinese cable is at any given moment 3) for all i know, pl2303 will not be gettable next month, next yr etc
mircea_popescu: so, tmsr-kernel, such as for instance the very tight musl stuff, or what cuntoo evnetually will become, is out of scope for this discussion. of course it should have native fg handling, and im sure it will.
asciilifeform: my whole point was that there can be no such thing as 'native fg handling' on pc. there's always a piece of shit in the way, e.g. usb.
mircea_popescu: neverthless, inca-kernel, be it "debian" or whatever it is, needs a way to be fucked such that it stops exposing any /dev/random AT ALL for as long as it is not exposing a fg random.
asciilifeform: aaa
asciilifeform: this is pretty easy.
mircea_popescu: that's all that's contemplated here, all we really want from the rest of the shitpile is a very clear knob, which, when turned, makes everything stop working that doesn't work like we expect it to work.
mircea_popescu: that's it and that's all.
mircea_popescu: and let them pick the pieces.
asciilifeform: trivially, the 'random.c' item linked earlier, simply cut it.
mircea_popescu: so you can waltz into dc installed "linux", patch it, and it either has /dev/random perma-blocked or else fg-streaming.
asciilifeform: ( observe, linus hasn't really got a kernel, just a tall pile of items like the one linked )
mircea_popescu: asciilifeform, well, ideally patch random.c into a fg streamer.
mircea_popescu: asciilifeform, quite.
asciilifeform: incidentally , knob won't 'break errything that isn't proper', troo champions of idiocy like gpg , will chug along without a working /dev/random ( iirc -- silently )
mircea_popescu: yawell.
asciilifeform puts the postgres docs down, drained for nao, goes back to bed, apparently did in fact pick up some sort of slow-burning infectious nasty on the return plane
mod6: ugh. get some rest, Sir
trinque: perhaps there's a yet more immediate route. /dev manager (eudev, w/e) is simply told to symlink /dev/random to the FG device.
mircea_popescu: trinque, that does at least half the job -- will get some actual entropy in there, even if it doesn't prevent the dilution with cvasi-random crap
lobbes: phew. I finally got logbot genesis up and inserting log lines into a database on my test machine. Hardest thing was figuring out the config knobs for postgres (and side-quests such as libuuid)
lobbes: Once I migrate this to pizarro shell, I plan to harvest my notes for a detailed n00b guide on standing up a 'vanilla' logbot (with focus on operating out of /home for pizarro shellists)
mircea_popescu: nice!
lobbes: I appreciate the suggested course correction earlier mircea_popescu. Now, after a week or so of toiling, I come out of the experience with more knowledge about V, sbcl, and postgres instead of painfully acquired trivia about dead-end things such as 'tcl' or the not-long-for-this-world gribble
mircea_popescu: word.
mircea_popescu: diana_coman, ahh it's nice to be back to braiding cordage...
ben_vulpes: mircea_popescu: ssh pubkey pls
mircea_popescu: use diana_coman's.
ben_vulpes: diana_coman: ssh pubkey pls
mircea_popescu: you already set a server for her yes ?
mircea_popescu: in other lulz : samuel l jackson is billed as "black guy" in http://trilema.com/2009/sea-of-love/
ben_vulpes: holy doodle i think that i do
ben_vulpes: notes ftw
mircea_popescu: the whole wot-isp thing is already paying off.
mircea_popescu: this is what i fucking wanted from the get-go.
mircea_popescu: https://www.youtube.com/watch?v=_T8PHEAHOBs
ben_vulpes: diana_coman: http://p.bvulpes.com/pastes/rRa4q/?raw=true
ben_vulpes bbl
ben_vulpes: my body is reacting very poorly to spring this year
diana_coman: ben_vulpes, ack & confirmed I'm in
ave1: http://btcbase.org/log/2018-04-23#1804661, can confirm arrival last week, I unpacked it yesterday, all layers seem to have been intact / untampered with! ☝︎
a111: Logged on 2018-04-23 00:35 asciilifeform: ave1 didja get it ?
mircea_popescu: ben_vulpes and it only goes downhill from here.
mircea_popescu: nice ave1 !
mircea_popescu: ben_vulpes, send me invoice for rest of year for it too.
mircea_popescu: incidentally, is your "standard" mpwp installation linkbroken by default ?
diana_coman: ben_vulpes, I don't need the wp stuff; what do I need to do to nuke it ?
mircea_popescu: rm -rs!!
diana_coman: lol
diana_coman: or uhm, ben_vulpes what's the link at which *my stuff* is served?
mircea_popescu: diana_coman, this usually goes the other way around : he tells you a ns, and i set the domain to point to it.
mircea_popescu: last time the process resembled pulling teeth, eventually ended up using qntra's ns.
mircea_popescu: they still have to create a zone for you.
diana_coman: I was hoping that part was already done :D
diana_coman: since I did not get any ns anyway
diana_coman: so basically now I can plonk all my stuff in there and ...wait
mircea_popescu: just about. you should be able to use the old reference style though, http://ip/~username/
diana_coman: apparently not: "not found"
diana_coman: I tried that first, earlier
mircea_popescu: i dunno wtf defaults they use.
diana_coman: hence my q to ben_vulpes as to what link I can use to see my stuff at all
mircea_popescu: word. ima bbl.
mod6: Mornin TMSR~
diana_coman: morning mod6 !
mircea_popescu: hola
mod6 waves
mod6: The weather has finally turned here! A few nice spring days we've had, after 6 months of winter.
mircea_popescu: nice then huh
mod6: Sure beats rolling blizzards, that's for sure.
mod6: Ugh, this coffee is horribru.
mod6: :D
mircea_popescu: aww.
mircea_popescu: life's too short for bad coffee.
mod6: yeah, I could probably degrease my lawn-mower engine with this rot.
mircea_popescu: why don't you come over ? i'll buy you a bag.
mod6: That'd be wonderful. I'd like to give real coffee a-go.
mod6: I saw on of your pictures (I think from a breakfast table), where it looked like there was a funnel/filter direct into the cup.
mod6: *one
mod6: http://trilema.com/wp-content/uploads/2018/04/iguana-1.jpg
mod6: Looks amazing!
mircea_popescu: yup.
mircea_popescu: they do not give a flying fuck over here, use 1/4 lb of fresh ground coffee for one cup, throw it away.
mod6: O.O
mircea_popescu: it's like it grows on trees.
mod6: lol
mircea_popescu: mod6, i kid you not, that's what you saw there.
mod6: Coffee opulence
BingoBoingo: diana_coman: Do you need any DNS records other than one pointing to your slice?
asciilifeform waves
mod6: Morning asciilifeform, feeling any better?
asciilifeform: sorta
BingoBoingo: Are your parotids normal sized?
asciilifeform: afaik
BingoBoingo: Probably not mumps then
mircea_popescu: BingoBoingo, yes lobbes needs logs.minigame.bz forwarded.
BingoBoingo: lobbes: Where do you need logs.minigame.biz pointed?
mircea_popescu: .bz
BingoBoingo: ty
lobbes: BingoBoingo: 192.121.170.137 plox
BingoBoingo: Aite, there is a zone on ns1.qntra.net and ns2.qntra.net for minigame.bz
mircea_popescu: diana_coman, say when ready i'll make the switch.
deedbot: http://thetarpit.org/posts/y04/070-bucharest-botanical-garden.html << The Tar Pit - Bucharest botanical garden in the spring, as viewed through a couple of camera lenses
diana_coman: mircea_popescu, I've transfered the stuff to the folder from ben_vulpes said apache serves; but more than that I can't check/see
mircea_popescu: aite let's see what happens if i plif the switch.
mircea_popescu: ugh somehow i fucked this up
mircea_popescu: BingoBoingo, it's .biz not .bz! sorry!
BingoBoingo: Aite, one sec
mircea_popescu: i have nfi how this is even fucking possible, but my dns zone is currently ns1 ns2.minigame.bz EXCEPT the domain is .biz
mircea_popescu: holy shit what the fuck is going on here.
mircea_popescu: oooh i see the problem. ok so, i owned both .biz and .bz, back when i used namecheap. i transferred the latter but not the former to namesilo back when all those shenanigans occured. so .bz item is captive in namecheap, Registry Expiry Date: 2019-07-01T16:40:17Z. but basically i'm letting it lapse ; so erryone will have to update to the .biz version and we can pretend this all never occured.
mircea_popescu: so BingoBoingo use .biz plox ; lobbes you'll be logs.minigame.biz rather than logs.minigame.bz as before. sorry about that. ☟︎
BingoBoingo: Aite, good to go
mircea_popescu: done here as well.
ben_vulpes: diana_coman: server is configured to respond to minigame.bz, per ^^ will be updating to serve minigame.biz and www.minigame.biz
ben_vulpes laboring under serious immune system load and cognitive deficit
diana_coman: ben_vulpes, k
ben_vulpes: diana_coman: everything in ~/www needs to be 0755, may i effect that for you?
trinque suggests something like diana_coman:apache and 0750
ben_vulpes: yeah?
ben_vulpes: i'm all ears, i've erred to date on what i think is safer but i lack a model
ben_vulpes: trinque: that makes total sense
trinque: if you do chmod g+rX it'll make directories traversable, but not go around marking everything executable ☟︎
trinque: can then do something like find /path/to/www -iname '*.php' -exec chmod g+x {} \;
ben_vulpes: right, neat
diana_coman: ben_vulpes, yes, go ahead
diana_coman: trinque's suggestions sounds good to me
lobbes: Aite. I'll have to make some minor tweaks later to reflect the new domain extension. Will let you know once logz are back in business >> http://btcbase.org/log/2018-04-23#1804804 ☝︎
a111: Logged on 2018-04-23 17:03 mircea_popescu: so BingoBoingo use .biz plox ; lobbes you'll be logs.minigame.biz rather than logs.minigame.bz as before. sorry about that.
ben_vulpes: diana_coman: website appears up to me, although that may just be /etc/hosts, i'm not up to speed on dns propagation yet
diana_coman: I can't yet see it from here - will give it some time and check again later
ben_vulpes: yup, just checked without local routing and no resolution.
diana_coman: btw thanks BingoBoingo
BingoBoingo: diana_coman: No problem at all
deedbot: http://qntra.net/2018/04/cisco-products-hit-with-vulnerability-in-firms-security-markup-language/ << Qntra - Cisco Products Hit With Vulnerability In Firm's "Security Markup Language"
asciilifeform: BingoBoingo: http://p.bvulpes.com/pastes/GzCyC/?raw=true
BingoBoingo: asciilifeform: ack
asciilifeform: BingoBoingo: ty
BingoBoingo: no, ty
mircea_popescu: http://btcbase.org/log/2018-04-23#1804815 << in this particular case it's nothing but a filestore, mostly for things like ... o hey look lobbes , eulora log no-go. ☝︎
a111: Logged on 2018-04-23 17:27 trinque: if you do chmod g+rX it'll make directories traversable, but not go around marking everything executable
mircea_popescu: annnd no, minigame.biz does not resolve. BingoBoingo ben_vulpes ^
mircea_popescu: you really need this to work easier than multi-day multi-attempt sorta thing.
mircea_popescu: "Begin testing delegation for minigame.biz. Name servers listed at parent: ns1.qntra.net,ns2.qntra.net Failed to find name servers of minigame.biz/IN. No name servers found at child." <
BingoBoingo: My bad, I forgot a line break answers now http://p.bvulpes.com/pastes/ME8bh/?raw=true
mircea_popescu: therew we go, foncirned.
mircea_popescu: shit internal references broken also. diana_coman when you have a moment run a sed /minigame.bz/minigame.biz/ plox.
diana_coman: ugh mircea_popescu I changed to minigame.biz and now the ownership of the files reverted to me and therefore index.html not available and it's again screwed; ben_vulpes ?
diana_coman: gah, I meant: permissions got screwed, not ownership
diana_coman: I'm not sure how this will work if groupownership of file changes from apache to my user every time I modify a file
diana_coman: files created in the ~/www dir should default /inherit apache group
mircea_popescu: diana_coman, you can just chown yourself neh
mircea_popescu: chown -r diana:apache or w/e
diana_coman: mircea_popescu, I need to change the *group*
mircea_popescu: was it apache:diana ?!
diana_coman: no,I need to change from diana:diana to diana:apache
diana_coman: hm
mircea_popescu: so do that.
mircea_popescu: but i'll admit this sounds somewhat awkward.
diana_coman: iirc that required root rights but will try
diana_coman: myah, operation not permitted
diana_coman: ben_vulpes, plox fix ^
mircea_popescu: well wait a second, wtf fix is this, "you can either edit your files or have them served by apache but not both" ?!
diana_coman: since it seems the fix to the fix is another wait, I'll set for now 755 on the files as otherwise ckang still won't be able to get the client and try the game
ben_vulpes: $user:$user and 0755 is what's been most useful on other sites; i think that'd be best here as well
diana_coman: why exactly?
diana_coman: I think trinque's suggestion makes perfect sense, as long as new files get by default user:apache too
ben_vulpes: ok i'll look into how to make new files inherit from parent dirs
spyked: TIL: setgid (g+s) should do the trick.
mircea_popescu: fwiw, all trilema.com files are trilema:trilema and somehow(tm) work.
ben_vulpes: spyked: thx, now to figure out what needs installing from portage to get setgid
ben_vulpes: ah i think i see how to get it with chmod
asciilifeform: mircea_popescu: is there a working mpex url ? i am standing snsa www / phuctor up today
asciilifeform: and the old one (.biz) dunwork
mircea_popescu: nope not yet.
mircea_popescu: but will be same one, so don't worry about it.
asciilifeform: ok ty
ben_vulpes: okay, gotit. diana_coman i'm going to run chmod -R g+srX your www dir
diana_coman: ben_vulpes, ok
ben_vulpes: please confirm that new files show up as diana_coman:apache
diana_coman: ben_vulpes, new files show up as diana_coman:apache BUT apparently still not with the right permissions to get served
diana_coman: at least I can change the permissions though to what they need
ben_vulpes: well that's still not great; they're not inheriting from the parent directory?
diana_coman: the file inherited the group ownership but not the group permissions from what I can tell
trinque: eh actually there were cobwebs between my ears when I gave that suggestion
ben_vulpes: i suppose they wouldn't, that being controlled by umask
trinque: probably easier move is to just add apache to the diana group
diana_coman: i.e. new file had r and write for user but nothing for group
ben_vulpes: diana_coman: yeah i think that comes from the user-level umask, which as i understand it sets permissions on individual files on create
deedbot: http://www.loper-os.org/?p=2277 << Loper OS - The Return of Phuctor!
asciilifeform: snsa www also updated.
asciilifeform: folx with snsa links to archive.is -- feel free to switch back to the realthing.
BingoBoingo: asciilifeform diana_coman Congrats
asciilifeform: ty BingoBoingo
asciilifeform: nao with ~100x faster db
asciilifeform: ( combo of ssd and various postgresisms )
ben_vulpes: i am interested in the postgrestisms you used
ben_vulpes: ah i actually had a conversation with a little lady about the 'dirty read' thing last night
asciilifeform: ben_vulpes: wouldja believe, nothing exotic
asciilifeform: ... indices.
asciilifeform: i had nfi they existed.
ben_vulpes: asciilifeform: lolk
asciilifeform: book-larnin'(tm)(r) ☟︎
ben_vulpes: megawatt of iron or two indices hue
asciilifeform: i didn't go to no db school, lel, ben_vulpes
asciilifeform: at any rate -- it worx.
trinque chortles
asciilifeform: chortle, it's fair.
trinque: naw, congrats on phuctor's rebirth!
asciilifeform: ty trinque !
asciilifeform: trinque: lemme know when you've added the rss back in; i'ma start up the werker later tonight
asciilifeform bbl, meat
mircea_popescu: asciilifeform, o hey!
ben_vulpes only ever attended cs program of hard knocks
mircea_popescu: http://phuctor.nosuchlabs.com/ << sweet.
mircea_popescu: now let's test the keys of folk who showed up in the meanwhile huh.
diana_coman: ben_vulpes, trinque this is just too brittle : how about moving dirs/files in there?
trinque: did ya see my last?
diana_coman: so I guess it'll have to be 755
mircea_popescu: !!key zx2c4 ☟︎
deedbot: http://wot.deedbot.org/AB9942E6D4A4CFC3412620A749FC7012A5DE03AE.asc
mircea_popescu: http://phuctor.nosuchlabs.com/gpgkey/8EEE946DA699B07DBC900780CEAF6AF1D30A962583586D3372E6C0C30688846C << ha-HA!
ben_vulpes: diana_coman: shall i add apache to the diana_coman group before you 755 up?
ben_vulpes sitting for remedial unix permissionsology at SOHN this afternoon
mircea_popescu: asciilifeform> i had nfi they existed. << well they were mentioned duyring the 3-4 optimization threads about 3-5x each time, but...
asciilifeform: mircea_popescu: funnily enuff still, i laboured under the misconceived notion that they existed ~by default~
mircea_popescu: do what indices i mean!
asciilifeform: didn't occur to me for even a moment that it doesn't make a btree for all params!!
asciilifeform: believe or not
asciilifeform: had to read the fucking docs, to learn this.
mircea_popescu: anuyway, i can see the angle ; but it's one of those cases where ai is much cheaper than just letting the person make the settings.
mircea_popescu: they take space, see ? lots of space.
ben_vulpes: \dt woulda shown you indices on the table too asciilifeform
mircea_popescu: some of my heavy hit tables are 60% indexes by weight.
asciilifeform: mircea_popescu: i also laboured under the notion that the thing 1) auto-indexed 2) to fill available space
mircea_popescu: yawell. different decade.
asciilifeform: (2) was suggested by all of the various lit suggesting 'expand room given for indices'
diana_coman: ben_vulpes, no, enough experimenting I'd say
ben_vulpes: thanks for your patience diana_coman
asciilifeform: ok i'ma genuinely bbl