#asciilifeform | 2019-12-06

the latter operates via heathen chats ?

14:53

Nah most are some kind of weird where you use "platform token" or OP_CHECKLOCKTIME_VERIFY to make trades.

14:53

i recall using some heathen 'shitc in, btc out' thing for squeezing juice from a forkcoin coupla yrs ago, but iirc that one vanished shortly after

14:55

But as far as coin things go, jfw 's scheme wallet most interesting item I've seen in year+ or more.

14:56

shinohai: it's the most readable such item i've yet encountered

14:56

even his orig. py prototype is quite readable.

14:57

realized, when reading, that i'ma need extended-euclidean in ffa ( and not only for eccisms, but e.g. cramer-shoup )

14:57

Dunno why I have such a distaste for py things. When you released shiva, I was pretty excited about using it for rpc things, but I'm not well versed in tinyscheme yet.

14:59

shinohai: as you know, i've written kilometres of py. e.g. device diddles, phuctor frontend, wartime logger .

14:59

it's 'a perl' tho, not fit for safety-critical systems, imho obvious why

15:01

*

asciilifeform orig. learned py to write plugins for 'ida', many yrs ago

15:02

It has it's places I guess, and of course ran jhvh1 fine for years. But as you stated above, certainly not something I'd want to build Bitcoin wallet on.

15:02

has same place as perl 20y ago.

15:02

'duct tape.'

15:04

bash, sed, and awk have always been my duct tape. Still use it to generate www pages xD

15:06

if there were a full-featured scripting lang made by civilized folk, would use that, and throw py out.

15:07

presently none exists tho. ( recall the convos re 'in what log bot?' )

15:09

Wasn't someone next door working an Adalisp or somesuch at one point?

15:09

i admit, never was able to understand the php aficionados -- what's the use of having a second, even uglier python where cannot e.g. persistent processes , or quick runs from shell ?

15:09

shinohai: spyked iirc made attempt

15:09

but it's larger proj than merely 'port tinyscheme to ada', need various knobs (sockets, db, at the very least) to be of practical use imho

15:10

I share distaste for php, why I never liked/used turdpress

15:11

I'm curious, what is asciilifeform 's favourite db ?

15:11

shinohai: i still have www w/ php (sewn from even older wp than mp's) but erry time i open the hood on it, is approx as pleasant as when fiddling old fortranisms

15:11

lol

15:11

shinohai: for phuctor and logotron used 'postgres' . would not describe as 'favourite', simply was best horse i found in that glue factory.

15:12

lone php script on my www atm is paste mechanism

15:12

It's like 50 loc so can stomach it.

15:14

for 'postgres' i at least know how to massage e.g. indices, so can be queried in not o(N). i expect this can be done for other sqltrons, but never found a persuasive reason to try.

15:14

imho best db for 'serious' proggy is an actual purpose-made data structure, with mmap to disk.

15:16

sqlism for db is rather similar to e.g. plasticine as construction material. great if you aren't certain what the shape is to be, but otherwise in erry way suboptimal.

15:18

*

shinohai ducks because using sqlite with bot ....

15:18

shinohai: the 1st phuctor stood on sqlite.

15:18

eventually led to minute+ loads.

15:19

i expect a log bot could run for 1000+ yrs on sqlite easily

15:19

It's pretty neat if not constantly under load, but yeah can see phuctor exhausting it.

15:20

iirc phf actually got away with having entire working set in ram, no db at all

15:20

( he never published a lick of src, all i've to go on is his commentary in #t at various times )

15:21

phf's front end to logger was always my favorite, that's one item I wish he had published. (along with patch viewer)

15:22

would've been great imho to have whole thing

15:22

instead of the current perlisms.

15:22

the patch viewer i'ma have to reimplement (w/ hopper for in-wot sigs) unless he somehow comes back to life

15:23

doesn't seem like anyone else is about to do it; and imho the 'oh hey i signed your proggy, nao where do i put this' situation is ridiculous

15:24

there's no reason for valid in-wot sig to languish as a wp comment until manually copied to 'code shelf' or the like

15:24

it's an eminently automatable thing

15:24

I've been busy past few months trying to organise all the patches/seals for stuff I use in single place. So have, e.g. trb ffa

15:24

asciilifeform: fwiw I don't see that problem; what I sign I put on my code shelf where it belongs.

15:25

sure, can automate the process further.

15:25

diana_coman: i'd rather prefer one where patch viewer and folx can upload sigs which then go where ought to

15:26

maybe; I don't really know why would I go about uploading my sig to someone else's site.

15:26

diana_coman: would upload to own site, primarily. and viewer can sync from other folxs' .

15:27

diana_coman: recall mp's crit of asciilifeform's 'i convert and upload jpg by hand', where he 'if i did this, would be 10x fewer photo tours, couldn't be arsed'

15:28

myeah, except here the bottleneck is people signing, not "too many sigs"; arguably at some point we get to too many sigs but I don't see it there yet.

15:29

diana_coman: the other reason for wwwistic patch viewer, is so as to link to lines of proggy. presently i have an extremely ugly hack for this.

15:29

( gnat's html gen thing doesn't in fact permit links to indiv. lines, but only (some!? why? nfi) subroutines; and of course entirely unaware of 'v' )

15:30

ideally, could link to any line in any vtronic proggy, and would be visible 'which patch', 'who signed' .

15:30

phf went most of the way to this before ran outta juice.

15:31

yeah, the problem that is easier to solve (for bonus points it also allows for a very elegant solution, yay); true.

15:31

*

asciilifeform actually started work on this in 2015, shortly after published orig. vtron -- but shelved, as phf , much handier wwwtronicist, put one up 1st. but phf never gave src.

15:33

diana_coman: would also like to be able to ~search~ the 'v yggdrasil' (to crib diana_coman's terminology)

15:34

I'd like first to have yggradsil.

15:34

( phf -- did not live long enuff to attempt this )

15:36

diana_coman: i.e. whole universe in one vtree ? iirc this was discussed even in '15. idea theoretically appealing, but i suspect in practice would break backs. picture if any time you published, had to 1st say 'pause' to 900 folx, or get 'orphaned'.

15:36

1990s versionatrons (e.g. 'cvs' if you recall) actually worked like this.

15:37

asciilifeform: interconnected/linked does not mean at all pause to everyone; but there's no imminent danger of that so not much point worrying about it now.

15:37

diana_coman: well, 'pause publication' rather than 'pause all work'. but afaik yes it would, work rather like btc chain.

15:38

if you were to have per-project rather than global manifest , tho, then -- not.

15:40

if folx willing to put up with having to regrind patch 3+ times -- as the folx who helped to fix e.g. asciilifeform's logger were -- then can do just about anyffin w/ v as-is. even 'whole universe' tree.

15:42

*

asciilifeform admires the pain tolerance of the folx who used, use, v. when i created it, half-expected that it'd collect dust, that folx would chicken out.

15:43

*

asciilifeform brb,tea

~ 1 hours 35 minutes ~

17:18

unrelatedly: folx interested in smmism (smm was how boobytraps were often implemented prior to proper fritzchip) will find that it is detailed in amd opteron datashit; and that you can singlestep smm handler w/ 'sage' probe on irons where there is connector for same.

17:18

snsabot

(trilema) 2019-12-06 dorion_road: In my brush up of, now wtf again is SMM ? this morning, I referred to an old article by robert. r. collins on the topic. He's new to me, but CV looks impressive, anyone here come across him ?

17:19

arguably not so interesting given as 'coreboot' traditionally has simply 'ret' in its smm handler; and afaik on all opterons where can connect 'sage', can emplace 'coreboot' in place of heathen bios.

17:23

smm was the orig. 'ring -1' on x86, introduced by intel in 486. worx like ordinary nonmaskable interrupt, but registers saved to dedicated on-chip sram, rather than to where stack points.

17:25

1990s lappies, e.g. 'toshiba libretto', that had bios-powered 'suspend to disk', usually implemented via smm.

17:27

whole point was to ~not~ interact with os (would often have e.g. battery handler in there, which had to run even if winblowz has crashed etc)

17:40

*

asciilifeform vaguely suspects that smmism was the 1st Officially nsa-ordered annex to pc arch. was ideal for undetectable boobytrap , i.e. entirely undetectable and unmodifiable from os, provided victim did not have a hardware debug probe (for intel boxes, was 100k$+ item that plugged under cpu, cable as thick as man's arm )

17:45

smm handler lived in ram segment that , once 'lock bit' in bios was set, could be neither read nor written, incl. from ring0 .

17:46

in the 'transistor-impoverished' era before they could put dedicated spy core in cpu proper -- was 'state of art' re snoopware.

17:48

*

asciilifeform at one time, decade+ ago, tried to implement an 'above os' debugger thing as a smm extension. but broke teeth, q of 'how to trigger' proved gnarly, there's no 'fire smm' button handily provided on any known mobo, the trigger is usually wired to southbridge directly

17:49

2 basic ways to make smm extension on heathen bioses -- 1) patch the bios and reflash 2) 'option rom's on pci cards. raid controllers at one time used the 2nd variant.

~ 1 hours 12 minutes ~

19:01

feedbot

http://qntra.net/2019/12/oracle-sues-us-department-of-labor-over-existence-of-administrative-courts/ << Qntra -- Oracle Sues US Department Of Labor Over Existence Of "Administrative Courts"

~ 3 hours 23 minutes ~

22:25

^Srry for join/partage, working on "translate" module and bot refused to rehash.

22:25