174700+ entries in 0.111s
BingoBoingo was a bit disappointed,
this was merely
traditional church building kept as warehouse. Was expecting USian phenomenon where new construction church is actually build in
the manner of a warehouse. Corrugated steel and all.
☟︎ BingoBoingo: OMG Costa Rica has warehouse churches
too!!!
mircea_popescu: Framedragger above described mechanism will do
that for free, yeah, just gotta have
to care enough
to put a link in log.
Framedragger: (i
thought you maybe wanted
to check if archive.is
tampered with it; which yes we should be able
to do as well)
mircea_popescu: instead of
that sad fate, pile
them into monthl;y
tarballs, deed
the previous month's
tarball each
time a new month starts and lo! major piece of infrastructure.
Framedragger: i just wasn't sure if you were looking for a particular downloaded-from-earlier copy of
that zip file
mircea_popescu: Framedragger imagine you know, five years from now
the situation repeats. except
then
there's no
tardstalk anymore. but it's ok, we have
the archive link. o look, it's just a dangling link.
Framedragger: i'll do
the latter, as
to
the former, i'm afraid it doesn't have it as of yet
mircea_popescu: there's no way he's fixcing
that however, outside of making a whole new
tmsr
http phf: so
the utf-8 parser
takes anything in
the 0-127 range as is, so we definitely wouldn't have any "magic string" cases in
the happy path (the unhappy path being
that you can apparently specify file
type and encoding when posting
to an rfc2388 form, and
the server can happily decide
to
transcode it in whatever way it pleases)
mircea_popescu: i
tried for isntance feeding
the original makefile in line by line with my comparer
thing, but all
the lines match.
mircea_popescu: i
thought maybe it is
the case
that it
takes a 2 or 3 byte item which starts with valid ascii and
tries
to interpret it as some sort of unicode bs and in
the process ends up applying a
transform
to
the
text somehow. but unless we have
the actual magic string, we're not finding
this.
mircea_popescu: anyway, so far : it's not clear
that it actually does anything untoward
to ascii ; my
tests as
they were failed
to find an example. we did discover it replaces nonascii chars with some random gunk, which yes should be changed but not really much of a point or anything
phf: nor would my explorations ever corroborate
that
mircea_popescu: and i am not about
to see 500 or 5 million instances of
that morph into some kind of actual proof of anything.
mircea_popescu: fwiw i never did manage
to verify
the saga was anything more
than alf's extremely short fuse.
phf: but i figure i'll
try pushing a binary file
through (which wasn't
the spec anyway, just wanted
to see what actually happens)
phf: i mean, we've had very many
tests demonstrating it, part of ongoing "paster ate my characters" asciilifeform saga
mircea_popescu: possibly
there's something else afot
though, because
that wouldn't have afected a makefile, ie
the original problem is it.
phf: foo is a specially crafted file which is simply a sequence of bytes from 0
to 255
mircea_popescu: phf yes but does it ever collate into an unicode character or anything of
the sort ?
phf: so it shouldn't be loosing anything in 0-127 range, at least
the foo file comes
through clean. it is
trying
to parse
the input as unicode (and i'm pretty sure it's
the server doing it, because curl is sending it 8bit clean)
mircea_popescu: anyway, i'm calling it good before i fill
the guy;'s disk with random lines.
mircea_popescu: pretty sure it simply drops non-ascii chars silently. which is
the right
thing at
that.
☟︎ trinque: phf's "foo" breaks it,
though really, if
the
thing's an ascii paster, I dunno
that
the bug's anything but "fed uniturds, did not
tell user
to fuck off"
mircea_popescu: atm watching it scroll, however many 100s of lines in,
they all match.
mircea_popescu very much recommends
this
test : pick random
textfile, put it
through
the above code, see if you find a dismatching line.
trinque: -n do not output
the
trailing newline
mircea_popescu: (i don't expect it does, because look in 2nd paste.
the -=== bit is produced by echo.
trinque: just... does. and look,
they gave ya a "echo -n"
to undo it! choice
trinque: I
think echo adds a newline, for one
mircea_popescu: trinque i am actually running a
terst and it's pretty mindboggling.
trinque: seems it wasn't
the
tabs;
those came back
trinque: spent more
time freaking out
than producing a diff would've
taken.
mircea_popescu: asciilifeform best i can
tell, wotpaste simply drops nonascii chars. was something in
the makefile not ascii ?
BingoBoingo: mircea_popescu: Hard
to
tell if usg.pinoy or PRC.Chinoy
mircea_popescu: BingoBoingo usg.pinoy moved on
to whining about imminent usg defeat ?
a111: Logged on 2017-07-16 03:22
trinque: seems sorta like
there's a double encode happening.
ben_vulpes: i'm very recently hamstrung in degrees of freedom by hardware failures, so i won't be
terribly responsive for
the immediate future. working on it
tho.
ben_vulpes: thanks for
the debugging input phf, mircea_popescu
sina: alright ladies and gentlemen, off
to visit some friends and have some pho
sina: double encoding is
the best and least confusing kind of encoding
trinque: seems sorta like
there's a double encode happening.
☟︎ phf: i say it right
there! "cat foo | wotpaste"
mircea_popescu: put your
test lines in
the first file ; examine
the results.txt file for
the results.
phf: common lisp specification defines a memory safe language, yes, but in
the crevices of "undefined behavior" you can have memory related exceptions. for example you can define a function with (safety 0) optimization declaration, which will neglect
to check
the
type of a passed argument, which might (and often will) result in a memory exception.
mircea_popescu: no idea why i'm
taking partial code/functions and making assumptions. here :
mircea_popescu: phf it does,
the actual stuff you're sending. i imagine you feed it
to curl via something somewhere in
the call yes ?
phf: no, because --data-binary
takes an argument
phf: sorry,
the
two are not actual files, but hexdumps of
the files i'm
trying
to post/getting back
mircea_popescu: oh wait, you post
the first one and get back
the 2nd one ?
phf: i checked
them,
they look correct
mircea_popescu: (nfi why we're even
testing uploading binary
to wotpaste, as
that's not what it's for, but w/e.)
phf: it's not, --data-binary
takes <data> as an argument, at least according
to man
phf: do you have a full command with --data-binary
that you're using
to upload?
phf: since i'm posting
to a form
though, it must be rfc 2388 compliant. afaiu posting raw binary data wouldn't work, because server needs
to know which field
the data is associated with
phf: yeah, wotpaste does
the pasting, curl
then pulls it back. i wonder if curl download negotiates it up
to unicode