tmsr - Search: " T"

26600+ entries in 0.015s

billymg: the way i understand it, client makes a request to url?start=5&end=7, server pulls content from mysql and returns with <span class="highlight"></span> injected where start/end indicate

a111: Logged on 2019-07-19 21:25 mp_en_viaje: can has trilema.com/article-title/?start=5&end=7 OR trilema.com/article-title/#selection but not rly both.

billymg: http://btcbase.org/log/2019-07-19#1923822 << why is the #selection piece necessary for the server side version of this? ☝︎

asciilifeform: afaik nobody has troo solution of yet.

asciilifeform: mipsism is a half-solution, of sorts, in that can build a mipsistic gnat and then will execute anywhere you can port the 12kB emu to

asciilifeform: trinque: to expand -- currently gnat srcball aint a complete description of gnat. i.e. it demands existing gnat. iirc was described pedantically in old thrd.

trinque: I'd be surprised if that stone doesn't haul in all the rest

asciilifeform: but iirc this was before we even gnat'd

asciilifeform: iirc at one time mp suggested this, 'there oughta be ONE genesis' ☟︎

asciilifeform: trinque: the obv. toothbreaking stone is gnat.

asciilifeform: trinque: thinking about it, 1 possible answer to the puzzler is to make an entire orchestra as 1 vtree.

lobbesbot: BingoBoingo: The operation succeeded.

BingoBoingo: !Qlater tell diana_coman http://p.bvulpes.com/pastes/XRfoU/?raw=true

asciilifeform: is actually all i got atm in the mips thing -- busybox

trinque: upstack, trying to capture linux in this manner might've been a fool's errand. might be that a muntzed kernel and a busybox-like are a saner beachball to eat. ☟︎☟︎

trinque: it's shitty, lol, but what am I going to hold when I'm 40.

trinque: might as well say what I'm doing with my time lately, which is related. I'm horking down bitcoins by the $10k, what

trinque: asciilifeform: the point being yes, we can be rid of portage, but where are the steps between there and here.

trinque: see, I'm this perverse fellow that knows the scripture even.

trinque: this I think correctly runs afoul of mp_en_viaje's "there is only one fucking tree, and it does not loop"

asciilifeform: if can solve this cleanly, can then be rid of portage, even, in principle ☟︎

asciilifeform: each of which by all logic wants to be sep. vtree, but are dependent functionally

asciilifeform: imho this is the real open q in the thread -- how to represent relation b/w nominally separate proggies

trinque: why doesn't it today?

asciilifeform: trinque: how does 'regress infinitely' ? plox to elaborate

asciilifeform: there's no fucking point to that, if we keep genesis of orig ☟︎

asciilifeform: nao what i'd really like is to ditch the idjit unixdiff's reversible deletions

asciilifeform: it's the same thing, imho, the cost of being able to trace the history to birth of universe, is having to keep it around

asciilifeform: i meant the blox, trinque

trinque: asciilifeform: the vtree for bitcoin is not and never will be 300gb

trinque: rather than having a perforation point which can be torn off, preserving the references for whichever historian that cared, but *not importing the shitwad into the future indefinitely* was the point

asciilifeform: trinque: indeed does. bitter pill. but sorta like bitcoin requires that 300GB (an' counting) of liquishit, 4evah

a111: Logged on 2019-07-19 14:52 mp_en_viaje: http://btcbase.org/log/2019-07-19#1923552 << exactly. it is marginally better than merely dropping portage on their head ; and it is ideologically defensible because guess what, if they don't like the genesis they don;t have to either patch on it or press it.

trinque: http://btcbase.org/log/2019-07-19#1923622 << the whole rest of history then requires this multi-gb genesis long after the resultant item has been carved to however many mb ☝︎☟︎

trinque: and I don't fault mp_en_viaje his ideological rigor. not in the slightest.

asciilifeform: i.e. it's a theoretical (if somewhat burning) q, rather than 'ohnoez, trinque Did It All Wrong'

asciilifeform: trinque: iirc we don't yet even have a vdiff that'd eat 900M

trinque: at any rate no, I'm not doing a man-month of rework on the thing any time soon. not standing in the way of anyone else doing it either.

BingoBoingo: So, it appears Iran took two British ships and released one with a warning. ☟︎

trinque: at the very least I'd want to say "this is the one with kernel in it"

trinque: not having to be able to read vpatches is, lets say retrospectively declarative.

trinque: I started from an item that was 900mb and that was just the ebuild tree.

a111: Logged on 2019-07-19 12:51 asciilifeform: trinque: it's 'gordian knot' that has to be cut ~somehow~. consider, as of yet asciilifeform cannot vtronicize that kernel patch, cuz... no kernel genesis.

trinque: http://btcbase.org/log/2019-07-19#1923536 << yes, and "cuntoo" would've contained it if I took the wad-of-shit-in-genesis approach. ☝︎

asciilifeform: 'new shithouse will be built from bricks of the one we blew up'(tm)(r)(lenin)

mp_en_viaje: now ima tell someone "look what weird shit we found, USE IT" and they won't dare say "why"

mp_en_viaje: tjhis, incidentally, is EXACTLY how tower of shit was even built in the first palce

mp_en_viaje: ok this might be major breakthrough here

asciilifeform: ah so if the page renderer asks for url, it dun see foo, bar ?

mp_en_viaje: moreover, with prefix, what do you do with trilema.com/4 ? currently it (correctly) finds http://trilema.com/2012/40-vs-80/

asciilifeform: and use'em to bake highlight

asciilifeform: nao all you'd need is to pluck out that foo and bar

mp_en_viaje: asciilifeform, admire the weird : http://trilema.com/2016/44/pili and trilema.com/2016/44/32/21/11/pili/ respecrtively

asciilifeform: e.g. trilema.com/rubber/ducky/2018/pushing-the-soft-flesh-and-so-on-against-the-so-and-so-blades-of-the-immutable-machine-etcetera-second-installment/

mp_en_viaje: in fact, it might be the one thing on the web i most use.

mp_en_viaje: and i use this ~extensively~

mp_en_viaje: eg, trilema has implicit search, if you type in trilema.com/pus you instaget http://trilema.com/2018/pushing-the-soft-flesh-and-so-on-against-the-so-and-so-blades-of-the-immutable-machine-etcetera-second-installment/

asciilifeform: mp_en_viaje: what if sel operands were to go on ~left~ side ?

mp_en_viaje: in theory you could possibly htccess script-fu something like domain/a/b/c into domain?q=a&w=b&e=c ; HOWEVER, the problem is the domain space is currently narrow.

asciilifeform: pretty typical specimen of 'we ameritards, did it over friday beer an' pizza' kludge

asciilifeform: rright, just thinkin' about .htaccess makes asciilifeform's stomach go

mp_en_viaje: as things stand, the ~only~ way to have user think url is "trilema/pili" while url really is trilema?p=65 or w/e it truly is, is via htaccess, which is like javascript written by monkeys in 1980

mp_en_viaje: it does, at that.

asciilifeform: loox like finally a potential justification for spyked's thing

mp_en_viaje: i mean obviously one could modify apache to be less fucked in the head, but ...

mp_en_viaje: the ~only~ way to have that indirection layer, is to use htaccess

mp_en_viaje: well, here's the thing -- IF you move the "extract data from uri" into php, you lose... url replacement.

asciilifeform: the proposed format aint mathematically ambiguous, i can write it as regexp say. but whothefuckknows what the programmatic liquishit getting in the way might be.

asciilifeform does not know how, exactly, so will have to leave it there

asciilifeform: unsurprisingly they dunwork ~now~ , but could be made ?

mp_en_viaje: consider some simple examples : http://trilema.com/pili works, takes one to http://trilema.com/2016/pili/ which is the cannonical form. http://trilema.com/pil also works

asciilifeform tried several times, never got his htaccess thing to work correctly, hence why still has barbarian ?=xyz titles

mp_en_viaje: afaik the name's hacked into an article via htaccess, which is yet another pointless and poorly specified scripting lang atop the shit tower

asciilifeform: why would it matter what's after the 2nd / ?

asciilifeform: mp_en_viaje: why's that ? name is what's b/w the 1st 2 / , neh ?

mp_en_viaje: asciilifeform, that'll never work though, because obviously it'll try to find an article by that name

diana_coman gave all computers around work to do for the next hours so she goes happily to sleep.

asciilifeform: ( down side -- would need the backend to search with ignoring of spaces, commas, etc )

asciilifeform: upstack: imho best form would be 100% plaintext, sumthing like trilema.com/article-title/rubber/ducky/ . then! can also ~select~ by hand crank, without jsism, if want , on box w/out jsism

asciilifeform: the thing already cuts into /-separated chunks, tho, neh.

asciilifeform has nfi whether this is trivial to implement , or headache

mp_en_viaje: ah that's an idea huh.

asciilifeform: cut it up on serv side into operands, do Right Thing

asciilifeform: could in principle trilema.com/article-title/5/7/#selection ?

mp_en_viaje: can has trilema.com/article-title/?start=5&end=7 OR trilema.com/article-title/#selection but not rly both. ☟︎

mp_en_viaje: and there's also no way to have trilema.com/article-title/?start=5&end=7#selection

asciilifeform: or is this in re the internal representation

asciilifeform: i thought footnotes were html anchors

mp_en_viaje: in continuing sads : /me attempted to spec the server side php job ; and it can be done, roughly, there exist primitives for, eg, "insert this after nth word and that after mth word". however : a) this interferes with the footnotes, because footnotes exist as z (( b)) and not as displayed on page ; and b) there's no good way to # and ? at the same time because of the way trilema handles title-urls.

asciilifeform peeks at the nfs forum crapola; nao with 'nya-nya' from peanut gallery, bonus. ☝︎

asciilifeform: all of this may or may not be worth the effort. but perhaps worth to try.

asciilifeform: bonus, in machine log could see what readers like to select.

asciilifeform: ( would do roughly same thing as presently, but munge the url in browser so when loaded, gives server-end selectified copy )

asciilifeform: and correct, if server-end munge, you dun need the displayer js at all

asciilifeform: naturally not. you'd have to put in e.g. <span style="blah">....</span> or whatnot. and 'anchor' so thing scrolls to it.

mp_en_viaje: asciilifeform, of course, if one's doing server side no need to grep, count start, count end

mp_en_viaje: local "select" behaviour is always defined, but afaik there's no exposed "make it look selected" call

asciilifeform: i haven't yet found browser that doesn't render text colour when asked

asciilifeform: no more so than w/ js selectism neh

mp_en_viaje: i honestly like this better.