Show Idle (>14 d.) Chans


← 2022-10-01 | 2022-10-03 →
11:05 phf aww yeah, congratulations signpost
~ 2 hours 59 minutes ~
14:05 signpost thanks y'all.
~ 44 minutes ~
14:49 * asciilifeform meanwhile tried to embed a simple utf8ism in 'data types' section to illustrate 'UString'; thought '2min job', ended up in astonishing hell of 'latex w/ other than latin? lol. and here are over9000 piles of kludge which dunwork'
14:51 asciilifeform latest loldiscovery is that the macros '\begin{htmlonly}'/'\end{htmlonly}' in hevea (to fence off a kludge that only needed for html output from proper latex) dun actually work ~at all~ (at least on asciilifeform's install).
14:52 asciilifeform ( on actual latex apparently ~can~ insert orcograms with a lolkludge using ancient ibm encodings -- but hevea in fact eats utf. )
14:58 phf asciilifeform, tex can't utf, so any language extension like latex requires cludges. there's over9000 of those, but if you "just want to"™ then you can easily use any of the alternative tex engines, built with latex extension. probably out of the box you'll have xelatex or lualatex. both talk utf native
15:00 phf xetex and luatex pass triptest and are tex for all practical purposes (hence legally have "tex" in name), but are otherwise built with gigabytes of liquishit required to support "modern" environment
15:08 phf (and in before "it dun work!11", both of the engines, like all tex, use computer modern by default, so your extra unicode characters won't appear, since they are not in comuter modern. you can select whatever alternative font, including your system fonts, using https://ctan.org/pkg/fontspec)
15:10 phf (and in before "i'm asciilifeform and everything is shit" here's the minimal latex document that compiles with xelatex on my system http://paste.deedbot.org/?id=k7Y8 )
15:21 phf also, i don't know what you're smoking, but this works for me, re htmlonly, http://paste.deedbot.org/?id=dtpv
~ 2 hours 22 minutes ~
17:44 * asciilifeform after posting earlier, turned up some of the items phf mentioned, still fiddling; likely botched sumthing in own setup
17:48 asciilifeform tried 'luatex' (on asciilifeform's box, 0.76.0-2013... tried to upgrade -- and earlier, to inst. xelatex -- but sank in dep hell) and phf's example chokes w/ 'File `luaotfload.sty' not found.'
17:51 asciilifeform phf: re: heveaism in particular -- does this work as described in manual on your box ?
17:51 asciilifeform ( on asciilifeform's -- nope )
17:53 asciilifeform ( and neither does e.g. this, tho fails differently -- under hevea, neither block executes, while under standard latex, ~both~ appear to )
17:53 asciilifeform in the former example, neither executes
17:55 asciilifeform ftr the latextron asciilifeform had been using, was: 'pdfTeX 3.1415926-2.5-1.40.14 (TeX Live 2013)'.
17:56 * asciilifeform evidently doomed to 'upgrade' the thing, i.e. to blow a multiweek vacation on manually emplacing the deps or alt. to bake a toilet box for subj (ugh)
17:56 dulapbot (asciilifeform) 2022-07-28 asciilifeform: billymg: sadly asciilifeform never properly wrote recipe for xorg -- will say that it 1) muchly depends on the gpu backend yer using ( asciilifeform -- amd ) ; and 2) consisted of ~month of trial&error, manually emerging package by package with --no-deps and walking the barf tree
18:01 asciilifeform ^ will prolly have to be finessed via makefile kludge. puzzling that the knob described in manual dunwork tho
18:01 asciilifeform ( the 'htmlonly' thing )
~ 42 minutes ~
18:43 jonsykkel is being able to emerge xorg with 0problems on dulap the exeptional case?
~ 19 minutes ~
19:03 jonsykkel http://logs.bitdash.io/pest/2022-09-30#1013554 << a++ giraffics, much harvard very h-index
19:03 bitbot Logged on 2022-09-30 22:20:45 asciilifeform[jonsykkel]: apropos of nuffin -- baking illustrations for rewritten pest spec
19:03 jonsykkel loking 4ward to spec
~ 21 minutes ~
19:24 asciilifeform lolty jonsykkel
19:26 asciilifeform http://logs.bitdash.io/pest/2022-10-02#1013607 << wasn't xorg per se, but the gpuism cruft ( 'mesa' + the board-specific liquishit for same )
19:26 bitbot Logged on 2022-10-02 18:43:43 jonsykkel: is being able to emerge xorg with 0problems on dulap the exeptional case?
19:26 dulapbot (asciilifeform) 2021-11-07 asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2021-11-07#1063729 << i got 1 of those amd cad card things that doesn't 3d ~at all. but worx with 100% blobless driver, and pushes 3 x '4k', which was all i needed.
19:31 jonsykkel ah okey
~ 30 minutes ~
20:02 asciilifeform fwiw this method worx (but not if the '\newcommand' is given where 'foo' and 'bar'...) . suffices for asciilifeform's purposes
20:14 asciilifeform the complete horrid kludge ftr; worx w/ ye olde pdftex if 'OT2' encoding installed.
20:15 asciilifeform what a timesink.
~ 35 minutes ~
20:51 phf http://logs.bitdash.io/pest/2022-10-02#1013598 << my intuition for this approach is that it wouldn't work, because tex is not a real programming language, and latex is its ungodly spawn. if i were to write from scratch i'd go straight for your second solution, with ifthenelse
20:51 bitbot Logged on 2022-10-02 17:51:16 asciilifeform[5]: phf: re: heveaism in particular -- does this work as described in manual on your box ?
20:51 asciilifeform phf: yea apparently \newcommand is prohibited in conditional blocks (and this doc'd nowhere)
20:59 phf it's more of a consequence of the way latex does things. those begin\end blocks are "special" in the sense of "retarded", and i wouldn't really put anything in them beyond rendering and rendering related commands
20:59 phf i don't really know latex, but if you know tex and try and do things the tex way it's pretty much guaranteed to go all kinds of wonky in a latex document
20:59 phf we've had #t threads about it…
21:00 * asciilifeform recalls several
21:00 * asciilifeform lulzily, had been latexing on/off for ~20y and still feels rather clueless justabout errytime, 'wtf, this oughta work'
21:02 asciilifeform is indeed a 'from hell' atrocity, 'which errybody stuck with cuz 0 substitutes'(tm)
21:04 phf i'm particularly surprised that not a single project outside of core knuth's tex even attempted to use cweb or whatever. i mean, you'd think with a fanclub of that size, the inner tex sanctum would at least larp a project or two. nope. zero.
21:06 * asciilifeform not used cweb
21:07 asciilifeform pretty interesting tho
21:07 phf it's the C version of knuth's WEB toolkit, for literate programming
21:07 asciilifeform aha
21:08 phf knuth regularly puts out little hacks that he wrote in cweb, while exploring some idea or other
21:08 phf and nobody else of course
21:09 phf https://www-cs-faculty.stanford.edu/~knuth/programs/random-dfs-a.w november 2021
21:10 phf but my question is why are even old school tex companion programs, like dvi2ps, not written in cweb
21:12 phf and the corresponding pdf output http://glyf.org/tmp/random-dfs-a.pdf
21:13 phf but of course you can't use hevea (out of the box) to render random-dfs-a.tex as html, because it's not latex, but rather tex
~ 29 minutes ~
21:43 asciilifeform right
21:44 * asciilifeform at one pt took a look at subj when in a fever dream of 'wainot deadtree ffa' but not got far
21:45 phf i still sometimes dream of "1000 year documents" or someshit
21:45 phf but if you've ever tried building minimal tex, realize very fast that there's no such thing as minimal tex
21:45 asciilifeform perhaps moar achievable than '1000y reichs' but only by a notch
21:46 asciilifeform and indeed sadly no such animal as 'compact tex' afaik
21:47 asciilifeform iirc tex predates (or at any rate conceived of in parallel with) even postscript, and hence support for the latter already bloats the kit and poorly impedance-matched
21:48 asciilifeform (whereas the native 'dvi' output format is eaten by ~nuffin afaik)
21:49 asciilifeform ... other than the various converters, that is
21:52 * asciilifeform often wonders what sorta proprietary lolware exists for mathematical typesetting; or rather, whether any of it ever advanced beyond microshit ordeal where 'pluck a symbol from this-here multipage box with mouse errytime'
21:53 asciilifeform afaik today even schoolbooks baked w/ latex ( with megatonne of custom styling so that 'not obv latex' )
21:54 * asciilifeform seen a few that were obv baked w/ 'mathematica' tho
21:54 phf asciilifeform, the problem as far as bloat is not just that "postscript is big", it's the fact that there's constant ongoing vertical integration, so most tools exist as a subdirecotry of texlive multi-gb distribution
21:54 asciilifeform ^ afaik the only real competition
21:54 asciilifeform phf: yea
21:55 phf so for example dvips existed as its own tool in the early 2000s, was maybe couple of thousand lines of code. then "integrated with kpathsea", which is a monster of a library to do path searches
21:56 asciilifeform asciilifeform's orig. ancient makefile for latex in fact invoked 'dvips', found that it chokes on tikzisms
21:57 asciilifeform good % of latexism has this 'witchcraft' flavour, hence asciilifeform's reluctance to 'upgrade' any of the pieces, his install is from '13 in fact
21:57 phf latex is further of a mess
21:58 asciilifeform the 'original sin' of that mess imho
21:58 phf probably back when lamport wrote it in 89 or whatever, was nice and elegant, but now it's a monster
21:58 asciilifeform reminiscent of cpp , which also started as 'elegant preprocessor for c'
21:58 phf it's a quentessential "hack". sort of like if somebody were to write DERIVE in brainfuck, because reasons
21:59 asciilifeform ^, and that nobody ever attempted to rewrite because 'aint nobody got the time'
21:59 asciilifeform indeed phakt that no one will ever get tenure or promotion to lieutenant-general etc from 'write sane typesetter', given as 'hey there's latex'
22:00 * asciilifeform recalls bolix's document system, tho. 100% parallel universe, and imho notbad
22:01 phf moon technology
22:01 asciilifeform it'd take a bolix-style effort (and sane bedrock) to dislodge tex, asciilifeform suspects
22:02 asciilifeform i.e. aint aboutta happen
22:12 phf http://logs.bitdash.io/pest/2022-10-02#1013654 << but then eventually https://tug.org/TUGboat/tb38-2/tb119tolusis.pdf
22:12 bitbot Logged on 2022-10-02 21:55:38 phf[awt]: so for example dvips existed as its own tool in the early 2000s, was maybe couple of thousand lines of code. then "integrated with kpathsea", which is a monster of a library to do path searches
22:13 phf "We present two extensions to dvips. One allows flexible inclusion of bitmap images and was imple- mented on top of the FreeImage library. The second extension solves quite a long-standing task: adding OpenType font support to dvips. Our extended dvips, xdvipsk, goes the “LuaTEX way” in Open- Type font management: it w
22:13 phf orks on DVI files com- piled by LuaTEX and expects to find the necessary Unicode map files, obtained as by-products of the compilation. The providing of these map files is en- sured by a special LATEX package."
22:14 phf that's also the version of dvips that lives in tex-live distribution, and the one that you have on your system
~ 51 minutes ~
23:05 * asciilifeform still not grasps why fonts gotta exist in completely diff vector-graphical format from the lines an' graphs etc., other than 'historical reasons'(tm)
23:06 asciilifeform far as asciilifeform concerned, whole shebang oughta've been a henderson-style tower of sexpr from day1
23:07 asciilifeform then not even would matter if rendering to vga or printer etc. ( incidentally sony made this go entirely reasonably in '80s w/ their 'display postscript' x11 replacement fwiw )
23:09 asciilifeform as it is, for fonts alone we've postscript, truetype, opentype, and wtf else
23:10 asciilifeform ( iirc traceable to a early '80s patent wank featuring crapple against adobe . )
23:10 phf possibly because computers were a mistake
23:11 phf i mean t1 postscript fonts were basically tiny postscript programs for each letter to render a letter glyph, but even then they managed to fuck it up: the actual per-letter postscript is obfuscated through some very very simple cypher, because adobe was being adobe
23:12 asciilifeform [insert oblig link to locklin's 'planes before cad actually flew in ~6mo from chalkboard] (tho too lazy to unearth link)
23:13 asciilifeform phf: iirc they also introduced some kinda 'unprincipled exception' kludge to enable kerning/ligatures where reqs state
23:13 phf of course somebody (i forgot who, it's in the lores, either unix room people, or smalltalk people) reversed the t1 format and "cracked" their scheme in a weekend, but the artifact stayed
23:15 phf asciilifeform, tex is actually font agnostic by design, it lays out boxes of boxes, and it's the "printer" that fills out the smallest glyph boxes with their corresponding glyphs. so like .dvi doesn't have letters in it, but instead says something like "font1-glyph10" and it's e.g. xdvi or dvips that puts the glyphs in pla
23:15 phf ce
23:15 asciilifeform 'softwarization' of phototypesetter (previously the things had actual wheels w/ projection film per glyph) prolly made adobe-style copyrastage inevitable
23:16 asciilifeform iirc tex actually briefly pre-dates the bitmapization of photosetter, hence ^
23:16 asciilifeform (i.e. had to work with the film wheels kind)
23:17 asciilifeform and, as is typical, the birth defect perma-baked-in
23:19 phf yeah but how solve. "the whole spitoon" basically mandates fits in head, that is rewritable in its entirety on demand. was looking at oberon again, and was kind of amused that to this day the output of Project Oberon is a handful of source files, without a compiler or anything. you're supposed to bootstrap it by whatever m
23:19 phf eans necessary :>
23:19 phf and then fits in head becomes monster once you want "my dissertation is on influence of 16th century japanese poetry on 17th century arabic sufi writing"
23:20 * asciilifeform after reading wirth's b00k, tried one time 'can actually run oberon?', promptly and unsurprisingly broke teeth
23:21 asciilifeform http://logs.bitdash.io/pest/2022-10-02#1013680 << >> subj fwiw
23:21 bitbot Logged on 2022-10-02 23:12:22 asciilifeform[jonsykkel|deedbot|awt]: [insert oblig link to locklin's 'planes before cad actually flew in ~6mo from chalkboard] (tho too lazy to unearth link)
23:22 phf asciilifeform, i've ran it, by using somebody's emulator. see you're not supposed to "just run it". you're supposed to have understood the machine from the documentation, and then spent couple of months implemting the vm and the compiler on your host system, until you can bootstrap it from textual sources :D
23:22 asciilifeform well asciilifeform in fact succeeded in pulling this off w/ (much simpler!) mips
23:23 asciilifeform for anyffin heavier, rapidly devolves to 'lock me in a pyramid for 500yrs' bolix ressurrection
23:24 * asciilifeform also did not, must point out, have to bake c compiler for that mips from empty space, either..
23:24 asciilifeform if had to -- and were enuff of a nut to insist -- would likely still be at it today
23:26 phf afaiu oberon is singificantly simpler than c. the exercise seems to be an equivalent of "implement this subset of pascal, PL/0" but maybe 10x of that
23:26 asciilifeform oblig '1000yr machine' oldthrd btw.
23:26 dulapbot (asciilifeform) 2022-05-16 asciilifeform: Opcode Semantics
23:27 phf there's a lot of this 1000yr machine, that's distinctly not a c-machine in the wirth's world. those guys seem to have a common methadology, of how to .e.g write a vm or a parser or a whatever in a couple of hours.
23:28 phf so a student of wirth method could probably get an oberon going on any random metal in a matter of hours or days
23:28 asciilifeform imho the 'scheme-83' folx, or even chuck moore et al, better example of '1000y' fwiw
23:28 asciilifeform and prolly yes
23:29 phf i think wirth people are roughly equivalent. i mean they had p-machine, and the whole PL/0, PL/1, ... bootstrapping strategy decades and decades ago
23:31 phf pl/0 was a small pascal introduced in wirth's 1976 algorithms+data structures=programs, which was elsewhere refined to make it a bit more pragmatic. of course there's like 5 people who understand this stack at this point, because, like lisp, it's singificantly lore and groking of specific philosophy, and if i had another l
23:31 phf ifetime, i would…
23:33 asciilifeform asciilifeform's favoured ada subset is justabout ~pascal. but not esp. fond of it from aesthetic pov
23:34 asciilifeform thing is, just about anyffin ~consistent~ would beat livingshit outta the extant horrors.
23:34 asciilifeform even 'pascal machine'
23:35 * asciilifeform has (or had? maybe threw) buncha '80s crapple 'developer' lit, where ~all examples pascalistic. at one time was the party line there
23:35 phf too late! you need to 13GB of disk, and as much ram, to boot the winner of machine wars: THE JAVASCRIPT MACHINE
23:36 * asciilifeform nostalgiates on 'SWAG'
23:36 dulapbot (trilema) 2015-08-17 asciilifeform: http://www.retroarchive.org/swag << proper archive.
23:37 phf isn't your favorite DERIVE written in turbopascal? :>
23:38 asciilifeform phf: afaik the frontend was! (the engine was 'mulisp' + a compiled bytecode turd for same)
23:38 asciilifeform frontend 'typeset' (in ascii graphic) the maffs
23:39 phf everything else is of the devil
23:40 * asciilifeform to this day impressed w/ ye olde 'derive', did ~95% of what 'macsyma' did, and not needed multiyear grant at lulministry
23:42 asciilifeform ( still 'state of art' in opensores symbolicmaffs afaik )
23:42 dulapbot (trilema) 2017-01-04 asciilifeform: (state of the art in open sores world is still macsyma, circa 1975)
23:42 * asciilifeform at one time dug up what was left of macsyma and put to use in anger in a saltmine
23:43 phf maxima is still being actively updated, but i've not had a chance to see what manner of code that is
23:44 phf it doesn't look like purely maggot code, because it'll be meaty updates to like (defun back-propagate-bernoulli-m4 ...) and shit like that
23:44 asciilifeform the 1 asciilifeform unearthed was iirc circa ~1980 but commonlispized
23:44 phf aah
23:44 asciilifeform no 'skins' or anyffin
23:45 phf ah, yeah, the last mit release, before symbolics claimed the whole thing
23:45 * asciilifeform at the time was doin' time in a darpa-powered lulzmine where orders were to 'canhaz auto-determinator of algo complexity when algo given in such-such format'
23:46 asciilifeform ^ no prizes for pointing out 'bbut halting problem'
23:48 phf halting problem is overrated :>
23:49 asciilifeform the thing, in turn, was outgrowth of a brain-melting pile of 'meta' bureaucracy, where 'ministry of lulz is swamped with claims of homomorphic cryptosystem; how to help evaluators evaluate'em'
23:49 phf one can still make some kind of useful claims even if you have "there be dragons" states
23:49 phf ahaha, that's beautiful
23:49 asciilifeform aaha
23:50 * asciilifeform did the obv thing and redefined the 'problem' into a 'solvable' one, then 'solved', fin
23:50 * asciilifeform ended up even sitting on committee, which was experience straight outta film 'brazil'
23:51 asciilifeform ( for the stars of the show -- perhaps from kafka )
23:51 phf http://logs.bitdash.io/trilema/2017-12-27#1759220 there is in fact a convergence point of the last few threads
23:51 bitbot (trilema) 2017-12-27 phf: early tex papers knuth talks about putting this or that formula into macsyma, but then from some point he fully switches to mathematica. breaks me heart
23:52 asciilifeform hm vol4 looks , fwiw, entirely texy
23:52 asciilifeform then again whoknows just with what baked
23:53 phf he's definitely using tex for taocp, but not macsyma anymore
23:53 asciilifeform a
23:53 phf in fact his "famous" fvwm screenshot got cli mathematica on it :(
23:53 asciilifeform ugh
23:53 * asciilifeform unsurprised tho
23:54 phf https://www-cs-faculty.stanford.edu/~knuth/screen.jpeg
23:54 asciilifeform lol a very vintage mathematica
23:55 phf probaly about as vintage as that screenshot ;)
23:55 asciilifeform ( was expecting it'd be at least of last ~decade, which nominally shits tex )
23:55 asciilifeform aa
23:55 * asciilifeform would also not be surprised if were ~current~ screenshit
23:56 asciilifeform 'openstep' & all
23:56 phf we've had a conversation where guy has been using a mac since forever
23:56 asciilifeform a hm.
23:57 * asciilifeform must bbl.
23:57 phf but also who knows, there's https://www-cs-faculty.stanford.edu/~knuth/zoomlunch.jpg but also https://www-cs-faculty.stanford.edu/~knuth/Knuth-vivian20181019C.jpg
23:58 phf that later 2018 one looks identical to the late 90s screenshot. and now we know
23:58 phf asciilifeform, laters!
← 2022-10-01 | 2022-10-03 →