279900+ entries in 0.173s
trinque: relatedly I'm about
to drop a vpatch for cmdbot
mircea_popescu: and a goat is closer
to a young woman
than any old whore - it's
tighter down
there.
trinque: in either case,
the
thing can be written such
that www is not a large part of what it is
mircea_popescu: asciilifeform once
there exist as many as five proper pages, you can have a proper browser.
trinque: it can be used as no more sophisticated a
thing
than a cmdline
tool
that shits
text
mircea_popescu: the big problem is
that lazy
thinkers and "hopefuls" of all stripes have made it into
their comfortable home.
this can't fucking stand.
mircea_popescu: in short :
the web is very, very broken - but most of
the breakage isn't even in
the software. yes apache sucks and wtf is js even.
trinque: I confess
to having already written parts of
the
thing.
mircea_popescu: a
tiny example, but indicative of
the conceptual difference.
mircea_popescu: but
the "media properties", from "one side"
to "the other same side" of "the spectrum" ... do not.
mircea_popescu: which is why
trilema, or qntra, have, say, a FUCKING ARCHIVE PAGE
mircea_popescu: trinque basically, it should be conceptualized not as a "this is a sheet i printed
this and
that on" but more like "what'd you like in your plate here ?"
trinque: mircea_popescu: yep,
thing could let
trusted folks write queries
mircea_popescu: whreas
the fucking internet is not made
to be a cable
tv substitute omfg. website should mean
that
the user can do a bunch of shit, according
to what he may need, some of which not necessarily foreseen by author.
mircea_popescu: trinque incidentally, i suspect
the entire website model is dumb as presented. currently, under pressure from ustards/business majors/other unwelcome masses, a website is a collection of webpages
that is very similar
to how a
tv show is a collection of
tv frames. MAYBE
they'll let
the user have a
tivo, but
that's at
the most.
trinque: yeah I can
take a crack at it. sounds like something
that has a list of SQL views, list of kinds of output. you pair
those up, allow filtering on fields (particularly
time)
mircea_popescu: shit, among all
this actual productive work i nearly missed
the lulz explosion on qntra.
trinque: it's generally useful for goddamned every web app
that ever webbed
mircea_popescu: asciilifeform honestly i was
thinking just pick one, clean it and repackage it.
trinque: I have need for
the reporting
thing myself anyway
mircea_popescu: if you build alf a web
tool
that looks and feels like whatever apple doohickey he fell in love with as a kid, he'll love you long
time.
mircea_popescu: trinque
the largest conundrum is
the
tex and sql end ; but
that can also wait.
mircea_popescu: certainly less wrong with openbsd
than... fucking red hat or w/e.
trinque: the missing
tool here sounds like a generic reporting engine
that speaks SQL and farts graphs, CSV
☟︎ trinque: plenty of lisp
tools
there
too; I've used hunchentoot plenty
trinque: openbsd's
httpd doesn't look like
that much src
to read, while we're still in c-machine hell.
mircea_popescu: myeah. but it's modular anyway, so. until such a
time, we just need a hm. either php or lisp i guess. or maybe hammer
the extant python-whatcha call it into shape.
mircea_popescu: anyway. in practice
there's also nginx, which i doubt is any better ; and i don't foresee our writing of a web server right nao.
☟︎ mircea_popescu: ya well. anyway,
this is kind-of
the direction i was dreaming for
the wp
thing.
mircea_popescu: this only partially solves
the www problem
though. servers have
to be lovingly helped
to stay up over ddos etc crapolade ; why
the fuck does apache not work properly etc etc. but
these we can abstract for now i hope.
mircea_popescu: for instance, it'd evident we want : a) csv, for
to grep ; b)
tmsr-html, which is NOT html, and not just because it does not allow js ; c)
tmsr-svg, which i'm going
to pretend is svg, because gnarly. somewhere inbetween b and c a latex fits in as previously discussed in logs.
mircea_popescu: it doesn't automatically have
to be "send your data
to X guy who runs
the official data-dump server"
mircea_popescu: anyway ;
the borders are fluid,
the "data dissemination" part can just live as a "this is
the official
tmsr web package ; install it like so ; it's maintained by
these people ; it does
these
things we want exactly right and no more"
mircea_popescu: at some point you will have
to like something ; lest something decides
to like you.
mircea_popescu: asciilifeform i'm sorry, were you saying "i am a flaming retard right now, so what i'm proposing is moving gb of crud around for no other purpose
than
to communicate a kb of information" ? i interpreted it
to mean "what if i actually have 1 gb of information, can i pass
that ?"
mircea_popescu: trilema serves many many
terrabites each month. what's a gigabyte.
mircea_popescu: asciilifeform so ? you push out csv's ; someone displays
them.
mircea_popescu: well
the alternative is me flying over
to washington with a rubber hose and pounding you into loving www.
mircea_popescu: i guess some work poured into a semblance of unified dissemination interface is unavoidable, hence
the shinohai wp project.
mircea_popescu: sort-of what's been happening with
the wot, for instance, anyway.
mircea_popescu: i'm
thinking
the way
this will end up is
to separate
the data processing ; data acquisition ; and data dissemination. evidently people good at any of
these aren't good at
the others, and have incredibly elaborate excuses for maintaining
the situation
that likely aren't worth cutting
through.
jurov: mircea_popescu: see later in
the log, apparently appended csv would enable corruption O.o
a111: Logged on 2016-07-18 17:09 jurov: and it's questionable why export it as html, when browser can't be used and we have
to grep it anyway?
a111: Logged on 2016-07-18 16:23 mircea_popescu: if you want a csv version, why not make
that. if you're making a html version, should have pages.
a111: Logged on 2016-07-18 16:49 jurov: heh, whois 108.163.248.235
times out
mircea_popescu: possibly found something like a nsa something or
the other, a botnet / spamnet etc.
a111: Logged on 2016-06-20 20:38 mircea_popescu: recall, jurov parsed all of github, produced a pile of keys and
the convertor code
a111: Logged on 2016-07-18 16:16 jurov: you need
to mitm someone logging in, no?
Framedragger: (hm, i'll go
thru
the ip clustering discussion again
tomorrow or
thereabouts, interesting stuff and interesting speculation how
this came
to be. need
to find
time
to do some actual anal-sis and graphing....)
Framedragger: asciilifeform: just curious, is all
that beefy server memory actually being used? by either djb gcd or db?
Framedragger: ah it's a separate
thing, right right. sorry i'm rambling without knowing / forgetting internals
Framedragger: effective_cache_size e.g. could be set
to 75% of
total system memory. otherwise - don't want
to presume - but it may end up paging
the fuck out out of
the disk. or other
things..
Framedragger: okay, but
there are other
things; but if it's
tuned
then fair enough
Framedragger: asciilifeform: postgres defaults are fucking stupid. `shared_buffers`, an important
thing when doing sorting, is 32 MiB at default