437400+ entries in 0.302s
cazalla: enjoys books
too which is good but reading jack and
the beanstalk for 20th
time each day gets a bit much, he's really picked up with
that past 1-2 months bring us books
to read him
cazalla: he's pretty clever, stacked a couple pillows in order
to get up on
the
tv unit
cazalla: gabriel_laddel, good but his
tantrums have kicked up a notch and he climbs everything
decimation: obviously somebody is pumping
these out by
the mega-boat load
decimation: yeah, also
there are probably hardware variants, according
to your research
decimation: sux.
they must have ripped out all
the debug for production cheapness
decimation: asciilifeform: re: solder balls: unfortunately sop
these days
trinque imagines cazalla
trying
to cross his eyes
to look at both
assbot: Logged on 20-07-2015 00:34:13; mircea_popescu: but past
that... who
the fuck cares.
trinque: gabriel_laddel: notably data modeling and analysis is one of
the Franz offerings
trinque: or why did I build
this cockpit for my business at all?
trinque: and I wanna goddamn *see*
the relationships
trinque: mhm, should be able
to add classifications at will all day long
ben_vulpes: coviariance analysis with shop humidity at
the
time
ben_vulpes: or perhaps we want
to
track quality ratings for *every widget produced*
gabriel_laddel: Allegrocache is
the only lisp solution
that will work for
this size dataset afaik, and I've spent a lot of
time looking.
phf: gribble returns identical strings for everyone, so
there's no way
to know if verify request is directed
to you or someone else
trinque: what are all
the classifications and distinctions involved in doing so?
trinque: where an index had
to be written for every "query" in advance
ben_vulpes: wow i just remembered
that gribble integration you wrote
trinque: but perhaps
that doesn't follow; I dunno yet
ben_vulpes: asciilifeform, phf, Adlai, gabriel_laddel: still curious about deploying CL code
to running instances
☟︎ trinque: I don't buy
that you should know in advance every interesting question you might ask your data
trinque: many user interfaces are essentially an editor for
the where clause, order by, and so on for some query
trinque: asciilifeform:
that doesn't solve it for me either, really
trinque: or
they happened upon a few
tools
that answered enough of
the essential questions
that
they never bothered asking more
trinque: most businesses out
there (that make a profit even!) are entirely blind
trinque: gabriel_laddel: I want
to sit down and catalog everything around me according
to kind and relationship
trinque: ben_vulpes: it's fine; eats a sql string or sexp version
thereof, farts list
gabriel_laddel: o anything complicated I simply ask for it. And I mean simply. I should never have
to put away
the
thing I'm working on unless I've actually finished it (fat chance say my publishers) or want
to do something else entirely."
gabriel_laddel: When I say 'work', I mean I want
to be able
to start
typing on
the screen, and if I feel like putting in a drawing, I draw on
the screen. Or I bring something from my scanner on
to
the screen, or I send something from my screen
to someone else. Or I get my Mac
to play
the
tune I've just written on
the screen on a synthesiser. Or well,
the list obviously is endless. And if I need any particular
tool
to enable me
to d
gabriel_laddel: 3. Have a bit of fun provided I've done enough of 2, which is rarely, but
that's another issue.
trinque: I'll
take a gun
to fire
tomorrow
too
ben_vulpes: there's a difference between chewing and a
tool
that abstracts a
thing
that needs doing
trinque: I am not sorry I used relational as a "gun
to fire
today"
trinque: asciilifeform: so am I meant
to read every damn object into memory just
to filter on >1 slot?
ben_vulpes: what about deploying code
to servers? just slime-connect
to
the remote host over an ssh
tunnel and
then compile
the new codebase in?
ben_vulpes: Adlai: once
told me
that
the 'log reading' period for CL was well in excess of
that for #b-a
ben_vulpes: i am going
to spend 2 years just reading CL
tooling documentation before putting anything into production at
this rate.
trinque: I have noticed many
times
that when someone
tries
to pry
the relational model from my hands, I lose behavior and am
then
told "you didn't actually need
that behavior"
phf: the loc on
that is in 10s, rather
then 1000s + external servers for when your first reaction is
to "reach for database"
phf: asciilifeform:
that's my answer
to.
the log is a contingency plan for when
the lisp instance fails, which it rarely does. in which case your goal is
to reconstruct
the state from
the
time of last save-lisp (say an hour),
till
the point of crash
trinque: I have
to say so far
the querying capacity of
this
thing looks
to be on par with couchdb
phf: in my experience it's cheaper
to literally go a log file and reconstruct data manually
the one
time your system crash,
then introduce uknowable redundancies
that
tend
to increase complexity and ultimately result in
the crash, because doesn't fit in head
ben_vulpes: you're
telling me
that i should what...just write code?
phf: i
think
the idea here is
that some data loss is way cheaper
then programmer
time. also memory is way cheaper
then programmer
time. if you have a really critical data stream, just do a write only log,
that you can either replay or even just recover manually
phf: i've crashed cmucl a few
times, but only when i would reach into heap
to access vectors directly. acl and lispworks never crash on me
decimation:
the cards very closely
to your breast, but if you need access, you just
decimation: you must lie, you are encouraged
to declare your private stuff and keep
decimation: " I actually
think C++ is ideal only for programmers without any ethics.
ben_vulpes: really? i'm supposed
to eat
this?
that lisp isntances don't crash?
ben_vulpes: or lisp instance crashign either,
though right?
ben_vulpes: slad doesn't really address
the power
toggle
thing
phf: yes, but without
the instance dying part
phf: my favorite way
to do lisp persistence is
to just keep everything in memory and do ext:save-lisp from a
that does minor amount of saved image management. i learned
the
trick from avi bryant back when he was writing interesting code
decimation: if you cannot predict
the outcome of a request of
the hardware, how can you possibly 'fix it' in software?
gabriel_laddel: haha, I believe no such
thing, but I have
to deliver "working" (for some value of
that word) software
to clients irrespective of how hardware behaves.
trinque: go kill a few of
those and fund your computer
trinque: old crufty industries
that are not interesting
to
those
trying
to advance
the art of computer science.
trinque: there are a hundred large pieces of software already out in
the wild begging
to be ripped open by something new
ben_vulpes: okay well
trinque gabriel_laddel i give up on
this lisp persistence
thing