Show Idle (>14 d.) Chans


← 2023-04-25 | 2023-04-27 →
awt[asciilifeform]: !!ticker btc usd
awt[asciilifeform]: We need online codes working so we can start sharing our uncucked LLM warez
awt[asciilifeform]: We can then train them on the logs and resurrect various historical republicans as bots.
asciilifeform distinctly recalls proposing an mpbot ( while the latter was alive ) but can't atm find in l0g
dulapbot: (trilema) 2016-09-14 asciilifeform: a slightly clever mpbot would be guided by what is actually said to it (grep the from:mircea_popescu lines for words or even fragments of the query and return shannonization of immediately subsequent from: s)
awt[asciilifeform]: asciilifeform: I've had some success querying chat gpt about trilema entries. Its immitation of his style was laughable though.
asciilifeform 'got enuff' shannonizer in 1990s; and not esp. impressed with the latest refinements to the concept fwiw
asciilifeform can see how these entertain folx who lived in a 'different 1990s' tho
awt[asciilifeform]: asciilifeform: I have found in most cases chat gpt at this point is far superior to googling for many programming tasks/errors.
awt[asciilifeform]: "How can I do x in y"
asciilifeform: awt: can easily picture (google has gone to utter shit)
unpx[asciilifeform]: I've been trying to understand this fancy computation they advertise in MAGMA but it still doesn't convince me a lot
unpx[asciilifeform]: awt, I'll wait to see it able to answer why that solution was chosen -- hoping it is not biased on ``it's fast''
awt[asciilifeform]: unpx: is this AI related? Not seeing the connection.
unpx[asciilifeform]: No, was just commenting what I've read. It is related to my math proj
asciilifeform: unpx: interesting piece, btw, asciilifeform read some yrs ago when was digging into algebratrons
asciilifeform: awt: loox like was redefined in '20 ?
awt[asciilifeform]: asciilifeform: yes. Looks like now missing IRA and Keogh balances. Dunno what Keogh balances are.
awt[asciilifeform]: However, is percent change, not total amount.
awt[asciilifeform]: So whatever the measure, continuously down only for some time.
awt[asciilifeform]: Possible that "Keogh" and IRA going up I guess. Mine isn't.
awt[asciilifeform]: Keogh is some kind of retirement plan for self-employed workers.
asciilifeform: ( ftr, asciilifeform's pov re reich stats. )
dulapbot: (asciilifeform) 2022-03-25 asciilifeform: the interesting, psychologically, aspect of the mats thread is the difficulty, for some folx, of making the jump to 'reich stats likely bogus, all i can rely on is personal experience of self and those around me'
dulapbot: (asciilifeform) 2021-12-29 asciilifeform: finds this experience surreal, in the old sovok nobody one'd be likely to meet entertained belief in Official stats
dulapbot: (asciilifeform) 2021-12-29 asciilifeform: the quickest way to see what reich stats are worth is to look at own wallet. asciilifeform's expenses in past 2y up considerably farther from the Official 6% or what was it.
awt[asciilifeform]: wen LLM ascii bot
asciilifeform: well for jobs like this, not even need mega-bot, but simply l0g search.
awt[asciilifeform]: "summarize asciilifeform's views on x" nevertheless would be nice
asciilifeform: awt: if you bake such a bot, dun hesitate to plug it in. oughta be at least good for some lulz
asciilifeform: ( and moar, for thread-completeness )
dulapbot: (asciilifeform) 2021-12-29 asciilifeform: 'avg. temperature in the hospital is only up 0.1 degree' 'but errybody is dying of typhus' 'so ok, we weigh the temp of the morgue occupants too, explain why shouldn't'
dulapbot: (asciilifeform) 2021-12-29 asciilifeform: 'let's include meth belt in house price index' yea
asciilifeform: http://logs.bitdash.io/pest/2023-04-26#1025630 << post some of these, for lulz, if still got'em
bitbot[asciilifeform]: Logged on 2023-04-26 10:58:19 awt: asciilifeform: I've had some success querying chat gpt about trilema entries. Its immitation of his style was laughable though.
asciilifeform: awt: ty
asciilifeform: awt: didja have to manually feed it the l0gz ? ( or already ate'em ? )
awt[asciilifeform]: Already eaten
asciilifeform: not even contains any obv howlers ( unlike the previous 1 iirc , where 'key developer of monero' or wat was it )
awt[asciilifeform]: Yes. Using a new version.
asciilifeform: 'Republic of Bitcoin (formerly the Most Serene Republic of Bitcoin)' << wonder what that came out of
asciilifeform wonders whether bot still generates the infamous great-looking lolcats but with odd #s of legs
asciilifeform: in wholly unrelated lulz, asciilifeform learned only nao that it is apparently entirely impossible to generate a unique function name using standard c macro.
asciilifeform: ( and even using nonstandard '__COUNTER__' apparently impossible to generate one that can be used twice+ )
asciilifeform: but if anyone knows otherwise -- plox to write in.
phf: maybe using __FILE__ __LINE__ ?
phf: or just __LINE__, won't be random but will be unique. main issue is that because everything is eval time, you can't capture __LINE__ in a macro, gotta register that function in a call from the same line
phf: 􏿽do i have an evil solution for you, you put X(foo) {...} in bar.c, and then in foo.c you #define X(a) int a##__LINE__() #include "bar.c" and elsewhere you do #define X register(a##__LINE__); #include "bar.c". because X(...) calls are all in the same lines in bar.c you're going
phf: 􏿽to get stable names across the different macro invocations
asciilifeform: wb phf !
asciilifeform: http://logs.bitdash.io/pest/2023-04-26#1025676 << possibly asciilifeform not read sufficiently carefully, but seems as if wouldn't work from inside anuther macro (which is where it was needed)
bitbot[asciilifeform]: Logged on 2023-04-26 15:35:26 phf[jonsykkel|deedbot|awt]: 􏿽do i have an evil solution for you, you put X(foo) {...} in bar.c, and then in foo.c you #define X(a) int a##__LINE__() #include "bar.c" and elsewhere you do #define X register(a##__LINE__); #include "bar.c". because X(...) calls are all in the same lines in
asciilifeform: ( full disclosure, what asciilifeform had was a lame thing where repl cmds defined macrologically, and wanted to abolish 'CMD' but still able to have e.g. '?' command which aint kosher in identifiers, so thought 'replace 'NAME' w/ gensym ?' but no trivial way to do it )
asciilifeform: as reader might've guessed, the exact value of 'repl_command_##NAME' -- unimportant, so long as globally unique under all known compiler & 'alignments of planets'
asciilifeform was rather surprised, after ~3 decades of suffering with cpp, that there's 0 clean traditional way to make this
asciilifeform: ...all the dross that's been added to cpp standard, but this -- nowhere to be found
asciilifeform: http://logs.bitdash.io/pest/2023-04-26#1025674 << was asciilifeform's 1st thought, but found no way to transform '__FILE__' into an identifier-kosher string in anyffin resembling a portable way
bitbot[asciilifeform]: Logged on 2023-04-26 15:25:10 phf[4]: maybe using __FILE__ __LINE__ ?
asciilifeform: ( and, interestingly, both __FILE__ and __LINE__ appear to choke clion's macroexpander thing )
phf: http://logs.bitdash.io/pest/2023-04-26#1025680 << well you gotta study the example in http://logs.bitdash.io/pest/2023-04-26#1025678 to understand how to solve
bitbot[asciilifeform]: Logged on 2023-04-26 16:50:49 asciilifeform[4]: http://logs.bitdash.io/pest/2023-04-26#1025676 << possibly asciilifeform not read sufficiently carefully, but seems as if wouldn't work from inside anuther macro (which is where it was needed)
bitbot[asciilifeform]: Logged on 2023-04-26 15:46:29 phf[4]: http://okturing.com/src/15646/body
phf: the idea is that you include the same file twice, with different definition of the macro
asciilifeform: if one could #define inside a #define (spoiler: mnope) this could be made edible. but as it is, lol
phf: well, it /works/, but i didn't promise it'll be elegant
asciilifeform: rright and a++ for 'obf c contest' but asciilifeform's orig. atrocity was strictly so to increase readability..
phf: in general though c requires some level of dry rejection. dry is gay anyway. void querycmd(char *q) ...; elsewhere registercmd("?", querycmd);
asciilifeform: well it's a cpp atrocity, so was able to banish the 'register' thing by baking an indexator thing
phf: i hear cpp11 or whatever has lambdas also
asciilifeform: ( used for knobs & repl cmds strictly )
asciilifeform: orig. wrote with lambdas, but cpp's lambdas behave peculiarly scopewise
phf: ah
asciilifeform: ( aaand, also wanted to be able to place knobs/cmds in arbitrary compilation units )
asciilifeform: ... rather than a central spittoon as seen damn near errywhere in similar proggies
phf[asciilifeform]: too much bookkeeping for my taste these days :>
asciilifeform 'damaged by lisp', developed long ago severe allergy to spurious entanglement b/w diff parts of proggy, and, in particular, anyffin resembling repetition ('table of knobs' a la trb etc)
asciilifeform: let the fucking machine make the tables.
phf: i'm not longer convinced of that principle. i mean we're of similar roots in this respect, but smh "machine making tables" turns into "programmer writes extra code"
asciilifeform: per asciilifeform's lights, so long as it's '1nce per lifetime extra coad' -- a++
phf: i don't understand that metric
asciilifeform: i.e. the ancient dream of reusable parts
asciilifeform: ( as illustrated there , nao if only the thing didn't require pointless but mandatory unique class name to be given )
bitbot[asciilifeform]: Logged on 2023-04-26 17:21:22 asciilifeform[4]: well it's a cpp atrocity, so was able to banish the 'register' thing by baking an indexator thing
asciilifeform: ^ allows a subclass of arbitrary class to be instantiated at runtime given associated name string
asciilifeform: ( cpp11, ftr )
phf: is this for your pest wunderwafffen?
asciilifeform: phf: it was in asciilifeform's attempt at a pestron yea
asciilifeform barely touched subj for several months, recently dusted off
asciilifeform got guism, 'console', uniturdism, and basic sqlite glue working; what remains -- well, the rest, lol
← 2023-04-25 | 2023-04-27 →