199900+ entries in 0.116s
trinque: asciilifeform: ~whole point~ is what
the implemented item would look like
trinque: phf:
the data is political??
trinque: that is
the standard, not some piece of software
trinque: over *massive* dataset which *is* relational regardless of
the db you choose
trinque: no, consider
the bundle of questions
that enterprise (oil firm, whatever) must ask itself on an ongoing basis
trinque: I would like
to hear of
the large firm
that did
their accounting on another
type of db.
phf: another alternative
to doing
the sql way is column stores (as an aside allegro cache is really more of a column store), where something like kdb is going
to be a reference (the apl approach
to databasing in general, of mmaping files with fixed size entries
that you can offset into).
phf: allegro cache is actually a lot closer
to how btcbase does it,
than
the postgresql way. internally
the
two are very different. if you
treat both as "a database" you're not going
to learn anything
phf: well, right, but you're not going
to learn how
to db by ~running~ databases.
the whole "db"
thing is an illusion anyway. rtrees, btrees, indexes, locking mechanisms, mvcc are all concrete algorithms,
that you can implement in an adhoc manner for your
task at hand and actually see how
they work and what
they do.
☟︎☟︎ trinque: I've said before
the
thing needs
to be split into many specific
tools
trinque: can't shitbag
the entire notion of "db" and expect
to do
that correctly
trinque: phf: I'm speaking from
the perspective
that
this data storage
thing eventually gets solved in
the republic.
phf: trinque: my point is ~why do you do what you do~.
there's no hard pressing imperative
to freak out, which is
the bezzle way of
things anyway.
Framedragger: "your
http client
tries
to re-request after request failed. DDoS!!"
Framedragger: but
that's
the last
time i brought
this up. i wanted
to clarify
that nobody's hammering it.
Framedragger: i'm running wget *by hand*. wtf. you expose a href on
the site. i
try
to get it. it fails. ?
Framedragger: after it gets a 504, it
tries again. sequential.
trinque: phf: you can avoid
the point as you choose. "there is a better land and we'll be
there someday; don't eat
til we get
there, best food on earth,
they have." has sunk many companies, and many people
a111: Logged on 2017-04-10 16:19 asciilifeform: and if folks insist on
trying
to bring it down, i will ban as much of ipv4 space as i have
to.
trinque: anyone who
thinks I'm defending SQL here is deaf
trinque: but we weren't
talking about
that; we were sidewinding onto whatever
trinque: for
that matter, your common lisp can't even dump a
textual representation of its entire state which can be loaded elsewhere disregarding arch
trinque: which is fine, but I assume you're not over
there running btcbase on allegro cache
trinque: none of
that is
trivial, and I'm aware of one mature such example in common lisp, and you have
to pay for it
trinque: I'm not going
to recite how mvcc works
to you, nor what atomicity is, you're being dishonest
trinque: phf: knowing how
to use it is
trivia, but
tell me, where can I put a dataset now, which I do under no circumstances want
to lose, which must be served concurrently
to a wide number of clients, which, and so on
trinque: I will wager one source of load is my RSS
thing, which you bitched wasn't producing results quickly enough
trinque: so
then your site is going
to be down in
teh meantime because of a combo of prior
technical decisions and present political ones
trinque: can either bitch like a woman or fix
the pain now and also move off later
☟︎ trinque: this conveniently missing
the point is shameful on you both.
phf: ascii actually can criticize postgresql from
the position of understanding. knowing intricacies of psql is not knowledge (rtrees, caching, etc.), it's
trivia ("you have
to
turn lever 3 and depress button Y"). ascii knows what he needs
to express, but he can't express it directly, because he's running against architectural constraints of his
tool. from
that perspective a general "databases are shit" is an entirely valid perspective.
☟︎ trinque: nobody's forcing him
to use it; now
that he has one, I've made a few suggestions
to lessen
the pain
trinque: I
think it's dishonorable
to prattle on in propaganda mode when
there are plenty of statements
to be made about why
the
thing's shit from
the position of understanding it.
phf: tmsr work is primarily defined by its voluntary nature, if i had
to do
things same way i do it at
the office i wouldn't bother. ascii doesn't know intricacies of psql from his day job, and i
think it's cruel and inhuman
to make him study psql ~as part of
tmsr work~. it's not
the kind of know how you get
to learn by sitting down with a cup of
tea and a large printout..
☟︎ trinque: yes phf;
this is
the only work I've ever done
phf: trinque is
the resident psql guru, he managed
to wire his 50 request lisp process
to a postgresql database
☟︎ trinque: there are many reasons
to hate rdbms. "cannot *at all* run something like phuctor" is not one of
them
trinque: are we doing
the "I used it wrong;
this proves my political views"
thread again?
BingoBoingo: asciilifeform:
Trilema is "common" application of single box. Many people
thusly blog. Phuctor is great because it is a big ball of weird RDBMS-BDSM
phf: i
take it phuctor is more like btcbase. a couple of
thousand requests a day, with an occasional massive bot spike
BingoBoingo: Phuctor, in addition
to.. you know its stated mission, is possibly
the greatest intelligence gathering exercise so far in "single box vs. hordes" problem
Framedragger: oh but like, *active* requests?
the use of
the word "zombie" implied
to me
that
the web requests
themselves may have already died. okok.
Framedragger: COPY (SELECT pid, client_addr, query_start, query FROM pg_stat_activity WHERE state <> 'active' ORDER BY client_addr ASC, query_start ASC)
TO '/tmp/pg-zombies.txt' WITH CSV DELIMITER ' ';
Framedragger: asciilifeform: does
the new-factor/key-adder write
to db under a different db user
than
the db reader which reads data for
the web backend?
Framedragger: have you checked if maybe google army is attacking you now
that you've changed
the robots policy?..
Framedragger: that's why i preluded with "if backend connects over
tcp"
ben_vulpes: asciilifeform: what was
the reasoning behind your call
that replacing
trb's boostisms with c++11isms was an assault on grandfathers pistols?
trb mustest compile with old gcc's?
ben_vulpes: mircea_popescu is in high spirits, must be nuts deep in something other
than cpp
ben_vulpes: whatever, does not matter in
the slightest.
a111: Logged on 2016-01-21 18:33 asciilifeform:
that is, using corrupted versions of gcc
ben_vulpes: what even is
the point of programming languages
that don't in2 closures
ben_vulpes: you know, before calling
the unary predicate
ben_vulpes: christ what even is
the point of a unary predicate in a language without closures
shinohai: To me, just easier
to grow damned spinach
BingoBoingo: After substantial effort
to boil out
the poisons. Ain't nobody got
time for
that.
shinohai: Older folks go nuts over
the shit here
though, eat it like spinach.