asciilifeform: it's ~10,000 man-years, in there.
asciilifeform: well i warned, it is satanic. but there is no human analogue of this tool, sadly.
mircea_popescu: they're certainly useful here.
mircea_popescu: yes but do i have to have mathematica now ?
mircea_popescu: yeah how fucking hard can this be, just take the ast and paint me some squares holy hell.
asciilifeform: i used to use 'mathematica' to do precisely this
mircea_popescu: i wish i had a way to program things in a visual manner, so i could obtain a GIF! animated! of what the algorithm is supposed to do.
mircea_popescu: im not sure the above notation is all that clear, but if any questions i'll gladly answer
mircea_popescu: last bit can in principle affect the whole damned ring.
mircea_popescu: nah, that's what my thing above : because of the state machine, you can't actually tell bit impact.
asciilifeform: it has another problem, that i never invented yet a pill against, which is that later bits in the turdogram 'count for' considerably less than earlier ones
mircea_popescu: still, it is ~only way to get hash that doesn't do the things we don't like.
a111: Logged on 2017-01-03 23:16 asciilifeform: to go back to hashes, and if you for some reason eschew 'when hiring fortune-teller, hire the cheapest',
mircea_popescu: it has the advantage that it needs an unspecified pile of memory (on average, half the message + half the message) for the state machine, and an unspecified number of operations (on average, 2x as many as message length).
mircea_popescu: . for each bit of M that is read : if 0 state machine gets a null bit added at the end ; if 1 state machine gets the M%S-th, 2M%S-th, ... nM%S-th bits flipped, for n=bitcount of S ; if 0 and the M-th % R-th bit = 1 then it is flipped, else it goes back to processing M-1th ; if 1 and the M-th % R-th bit = M-th % S-th bit then it is flipped, else it goes back to processing M-1th. that sort of thing.
mircea_popescu: asciilifeform btw it occurs to me that your proposed padding scheme, while not useful as a rsa padding scheme, nevertheless may be rescued into a very serviceable hash function, which has the notable property that a) acordion and b) user settable output size. something like this : let R be a ring buffer of as many bits as the user specified the output should be, let M be the message. let there be a one bit state machine S = 0
trinque: just another day on fleanode
mircea_popescu: BingoBoingo neah, the imaginary ethereum exchange rate went EVEN HIGHER THAN BTC!!111
BingoBoingo: And they work off of some poor folk voucher system instead of actual trb
BingoBoingo: Seriously how does she make that much? I just assumed the decimal place moved.
a111: Logged on 2014-10-08 19:19 mircea_popescu: einstein never got the damned formula out because luce irigaray didn't see why he'd privilege the speed of light over other speeds that are so much more important to us. and so on.
trinque: asciilifeform: they're telling you to do retarded things with your product to satisfy their own emotional needs
mod6: <+asciilifeform> except he dun like the auditability thing << wat
BingoBoingo: But yeah, didn't consider the construct before replying
BingoBoingo: Hussein bahamas fictions himself a "way the world works" and dooms plenty well
mircea_popescu: kinda the problem with fiction, either it's "doom" and then a simple word, or else an explanation, which necessarily cuts short.
BingoBoingo: There's plenty of doom in fiction
mircea_popescu: well there's no doom in fiction.
asciilifeform: prolly the latter, d00d is almost lethally lazy
asciilifeform: except he dun like the auditability thing
mircea_popescu: so all the better.
asciilifeform: in other lulz, herr kako loves us enough to try to make own FUCKGOATS and try to undercut
mircea_popescu: might haver changed in the meanwhile.
a111: Logged on 2017-01-04 13:18 Framedragger: (i.e., need to think carefully about the value of $time http://btcbase.org/log/2016-12-23#1588759 but may be useful)
asciilifeform: 'oh it's the same text except that i ran this-here perl turd on it and trust me that it works as i said on your perltron' is not vtronic.
a111: Logged on 2017-01-04 10:33 davout: i don't really see asciilifeform's issue with large 'formatting' patches, as long as it can be mechanically established that the changes a patch brings do not change any of the code semantics there should be no problem with arbitrarily large patches
mircea_popescu: and you JUST KNOW the conclusion of this will be "mp is bad and evil", in NO FUCKING CASE "i was an idiot and who knew, turns out it's unsustainable".
mircea_popescu: and in unrelated lulz, http://trilema.com/2015/internoc24-or-the-crisis-and-its-resolution/#comment-120934 "oh, lalala we can't hear anything" ; two months later "oh wait, what the guy said would happen happened, our marketing's utterly shot now" "OH I KNOW!!! I WILL FIX EVERYTHING! by going on in the manner that got me raped in the first place, because i'm a speshul snoflake that can!11".
davout: have trb rescan the UTXO for each "gimme-UTXOs-pertaining-to-these-addresses" and see how that goes
davout: and if history is somehow lost by the wallet, it's a rescan away
davout: if the responsibility of maintaning the address history is delegated to the wallet, i expect that this index can be massively shrunk
davout: no just the unspent ones
davout: for the cost of a 20gb index the wallet code can be completely removed, and implemented as a couple light scripts on top of TRB
mircea_popescu: that's rarer, most wallet users being tardstalkers the solution was "use website"
davout: so apparently, the electrum folks manage to fit a complete TXOs index in ~20gb
mircea_popescu: davout i meant as far as actual mining is concerned (in practice). by the time of the gpu era, most people ditched satoshi code for mining.
Framedragger: (i.e., need to think carefully about the value of $time http://btcbase.org/log/2016-12-23#1588759 but may be useful)
davout: i don't really see asciilifeform's issue with large 'formatting' patches, as long as it can be mechanically established that the changes a patch brings do not change any of the code semantics there should be no problem with arbitrarily large patches
davout: ah i didn't notice the two lines merged as one being seen as a move
ben_vulpes: if the dark green indicates "preserved" then it's confusing the opener for (format with the opener for (let
ben_vulpes: also the preserved section goes from one line to two, and the diff-o-tron isn't saying anything about the leading paren on that bottom line
ben_vulpes: notice how the mapcar line is highlight moar red to convey moar meaning
ben_vulpes: davout: i think the highlighting is trying to tell me that it thinks that `format t "~{~A~}"' persisted from the previous commit to this one
ben_vulpes: http://imgur.com/a/KuPJv << some kind of voodoo magic in there knows i moved that format string around
ben_vulpes: http://btcbase.org/log/2015-09-09#1268284 << for the life of me i cannot find the thread to which you're referring here
a111: Logged on 2017-01-04 02:55 asciilifeform: (possibly davout might have some input) i recently read an american-flavoured thing re pilotage accidents, and it dwelled on 'jp in petrol tank' , insidious condition where the engine ~will~ start but tends to quit ~during takeoff~, guaranteed corpse
davout: anyway, i think the benefits of extracting/dropping the embedded miner would be quite small in contrast with the benefits of untangling the wallet
trinque: probably referring to that most miners are using their own strange to mine
davout: http://btcbase.org/log/2017-01-04#1596028 <<< not in trb as far as i can tell, i can tell the thing to 'generate' etc.
mircea_popescu: anyway, i kind-of gave up documenting their downtime.
mircea_popescu: wasn;t there an automake for kubinetes or such ?
mircea_popescu: but to please ben_vulpes : it may amuse you to learn that ask.fm put in a mitigation for my bot (ill designed, and fail to work) sometime on dec 30th. AND THEN apparently (accidentally ?) reverted it sometime jan 2nd.
mircea_popescu: also is the first stone of "well your computer doesn't actuallty work, does it" most normies may encounter.
trinque: filters out some power rangers right there
mircea_popescu: precisely the "trivial" thus instructive sort of task
mircea_popescu: but anyway, this is actually a point - maintaining a stable irc connection is both informative and good preparation for the tasks ahead.
asciilifeform: i actually like the crc encyclopaedia moar
asciilifeform: ben_vulpes: the blue-cover crapola thing is technically a different b00k
ben_vulpes: these episode 2 crickets are unsettling, perhaps i need to write more things for people to disagree with
a111: Logged on 2017-01-04 03:14 mod6: ben_vulpes: fyi, second edition has the red cover, the first edition is the blue cover one.
mircea_popescu: you want it to go through a www page, find all signature blocks, and identify it as a fingerprint then ?
asciilifeform: a 'this www has apparent pgp signature, let's see whose, and for the record', was the idea.
mircea_popescu: this is so horribly stated. so what you want is, for trinque or you, these being the only "et al" curtrently keeping pgp data ; to implement a search through it by random string ?
asciilifeform: (btw, trinque et al, here's a bot command idea: #!k (for instance) takes url and looks for pgp keyblocks or sigs or any other pgptronic object from which a key bitness , fp, or other interesting attribute can be pgpdump'd, and prints same in log)
trinque: The state’s highest court explained, “the detective was the only eyewitness who testified to the defendant's conduct” and that he was disgusted “after viewing the defendant's exposed penis, not for himself, but rather out of ‘concern’ for the women seated on the bench.”
trinque: https://www.rt.com/usa/372609-public-decency-boston-law/ << ahahaha, hey mats, let it be known that you can jack it on the subway up there so long as nobody's "shocked" or "alarmed"
asciilifeform: (linked mainly for the comments. seriously, wtf)
asciilifeform: rather like real-life spacecraft.
a111: Logged on 2017-01-03 22:48 davout: asciilifeform: point is piloting a small plane there's just a few things to pay attention to constantly
mircea_popescu: http://btcbase.org/log/2017-01-03#1595972 << kinda where all those "great future everyone flies plane" threads always die, with mp saying "i wouldn't put that work in if the plane sucked my cock."
a111: Logged on 2017-01-03 22:35 asciilifeform: ben_vulpes: my current (wholly nonexpert) understanding is that airplane DELIBERATELY omits 'hard interlocks' wherever possible, on the principle that not-being-able-to-X-when-you-must is worse than can-X-when-you-mustn't
mircea_popescu: http://btcbase.org/log/2017-01-03#1595914 << this is just about it ; the python 3 cancer hasn't eaten through most of that yet. at least afaik.
mod6: yeah, can wreck the thing for sure.
mircea_popescu: mod6 theoretically cylinders blow.
mod6: yeah, actually, now that I think about it, i was told that in this particular case I was talking about, the guy didn't even put in regular unleaded (87 octane), he put in like 110 octane racing fuel.
asciilifeform: they have clever nozzles now, but there are always mega-heroes who manage to fuel up with entirely wrong liquid
asciilifeform: (possibly davout might have some input) i recently read an american-flavoured thing re pilotage accidents, and it dwelled on 'jp in petrol tank' , insidious condition where the engine ~will~ start but tends to quit ~during takeoff~, guaranteed corpse
mod6: know someone this happened to once, apparently diesel was shooting flames of unleaded out of the tailpile, while running like dog-shit (excessive knocking etc).
mircea_popescu: and in other news, girl preparing to weigh self, "if it shows me over what i started at ima jump out the window". me, equanimous "wouldn't it be better to jump out window if surprisingly light, than if surprisingly heavy ?'
asciilifeform: '... and that was how we found that the average policeman is not so smart, but very strong' (from -- iirc -- one of mircea_popescu's articles)
mod6: it easy to put unleaded regular gas into a diesel, but not vice-versa. the diesal nozzles are too large for a standard gasoline tube.
asciilifeform often wondered why the things can't simply be made threaded, and with variant thread
asciilifeform: mircea_popescu: in usa, the ~only way to achieve this 'feat' is to pour into jerrycan first
a111: Logged on 2017-01-03 22:32 asciilifeform: i'd naively think that this would've been resolved in 1930s, if not earlier, just make the levers vastly different (shapes, or lengths, and feel, etc)
mircea_popescu: in other news wow Framedragger your log selection thing is immensely useful for eg leaving comments on davout 's site.
mircea_popescu: (re the above discussion - one of the B.Ts may well be "cache of txns with unspent outputs" specifically to aid in mempool evaluation. but this is philosophy 102.)
mircea_popescu: (re the above discussion - one of the B.Ts may well be "cache of txns with unspent outputs" specifically to aid in mempool evaluation. but this is philosophy 102.)
mircea_popescu: these are good-to-have, not dependencies. let the lizzard queen fuck with ips a while first.
asciilifeform: you gotta have the crypto auth.
mircea_popescu: asciilifeform nobody forces you to keep the throttle in place for your friends.
mircea_popescu: this is a fine avenue of rebalancing the miner/node nonsense.
mircea_popescu: worked while it worked but on the long term "collegiallity" of this elk is unsustainable.
asciilifeform: mircea_popescu: here i gotta agree, the 'allcomers can get their tx candidates evaluated' is doomed
mircea_popescu: how MUCH verification work i wish to do for the world - is my choice, not the world's.
mircea_popescu: you cvan not engage in an open ended "i will for all comers do the work of checking one cent txn against a 10 dollar blockchain".
asciilifeform: (a tx can have valid form in all sense other than being a doublespend, mircea_popescu knows this)
mircea_popescu: then no wonder your node falls over.
asciilifeform: in the case of verifying a block -- it does
mircea_popescu: and reads the blockchain to do that ?
asciilifeform: well the old code has ban(...) which instructs 'this datum was a malicious turd and i dun want no moar from that ip'
asciilifeform: perhaps if nothing could possibly land in the queues faster than it can be eaten, this would not be so.
asciilifeform: while i like the 'diode' aspect of this scheme, i will say that my public nodes would fall down in about three seconds, and permanently, if the primitive autobanhammer of gaviniferously-symptomatic peers were to be switched off.
mircea_popescu: (the correct solving scheme is still as i said back when we were discussing mempools, to keep track of peers (yes, by ips) and score them by the fees they bring your mempool. with this change -- that is even implementable.)
mircea_popescu: and no, not wholly mempool less. there is m.t. what it contains - we care not. when problems will arise, they will be solved without impacting on the core scheme.
mircea_popescu: and in other alt-universes, http://68.media.tumblr.com/29a57cee1f8350a5c981f7ec5d318221/tumblr_mqko9plKqo1r7dgcpo8_1280.png
mircea_popescu: i don't think today's logic does anything ; and i don't expect carrying it forward is useful. spec does include room for trb.n to do some banning, including on the basis of passively exfiltrated data from trb.b. that a protocol for this purpose may later develop i don't dispute, but it's not included both because it's not needed and because it can't become a "dependency". it's not.
asciilifeform: and whatdoyoumean 'no tx', is the thing contemplated a wholly mempool-less node ?
asciilifeform: (banning would, as far as i can tell, require 2way comms b/w the modules)
asciilifeform: in the case of candidate-block and candidate-mempooltx queue receiver (from outside the walls), mircea_popescu proposes to throw out even the weak logic for banning obvious crapola that we have today in trb?
mircea_popescu: so then.
mircea_popescu: there is no tx. you do not read. what is N.B ?
asciilifeform: in the tx case, anybody can overwrite legit tx in your buffer by dumping liquishit in.
mircea_popescu: so then what happens ?
a111: Logged on 2017-01-03 20:48 mircea_popescu: in particular N.B should be "older overwrites newer" style ring buffer. of particular concern are situations where the buffer is set shorter than the longest reorg, in which case the node will wedge. TRB.N not accepting blocks with index lower than highest of B.B is for sure not feasible. "how many behind" should be an operator knob.
asciilifeform: what happens when evilpeer dumps a TB of liquishit into the queue?
mircea_popescu: no. one just reads, the other just writes, at all points where they interact. no talking is contemplated, and if this is "a protocol" then it's already given.
a111: Logged on 2017-01-03 22:24 asciilifeform: they will require a -- quite complicated -- entirely new protocol
mircea_popescu: http://btcbase.org/log/2017-01-03#1595883 << the only reason to even do it is if no protocol is created at all.
a111: Logged on 2017-01-03 22:19 davout: make the miner a separate bin
mircea_popescu: http://btcbase.org/log/2017-01-03#1595862 << this is a stronger argument than it appears. we may find ourselves in the position where we have to, if not "mine" in the current sense, say what mining should be. in no case can it be "oh, mining, not something we care about". about mining, about any other part.
jhvh1: 9. Made direct amends to such people wherever possible, except when to do so would injure them or others.
asciilifeform: (every tx in memory relies on its inputs being present , and at all times becomes a threat to crash the process if one should turn out not to be there. ~all pointers potentially dangle . thing is as rotten as could be imagined.)
asciilifeform: and there are no provisions for safely ~removing a tx~ from mempool, ever, at all
asciilifeform: and not only of std::map in the abstract, but of the same set of maps everywhere in trb, simultaneously, and at the same time with the pestilential global locks
a111: Logged on 2017-01-03 22:06 asciilifeform: asciilifeform's (and later again jurov's) utter failure to unravel the heap, at least, suggests this.
jurov: http://btcbase.org/log/2017-01-03#1595849 iirc the issue is pervasive use of std::map, which fucks with the heap like horny pig
asciilifeform: esp. if the actual computation is unbounded
asciilifeform: ben_vulpes: a non-algebraic ('programmatic') hash algo opens the possibility of crafted cpu-ddos
ben_vulpes: wyrdmantis: again with the laptop messages
asciilifeform: they are arithmetical, because the designers insisted that the hash be computable in fixed number of cpu cycles.
asciilifeform: essentially, anything where you cannot, in any practical computer, express the hash's reversal as an n-sat problem
asciilifeform: i know of 0 uses for a 'hash' where the same ~input~ is not guaranteed to produce ~same output~
asciilifeform: to go back to hashes, and if you for some reason eschew 'when hiring fortune-teller, hire the cheapest',
asciilifeform: ben_vulpes: if you know how to get effect entirely analogous to gas turbulence in a purely electric machine, there are many folks who will clap, do say.
asciilifeform: there is ~0 actual relationship between 'confusing to the naked eye' and 'crypto-hard'
asciilifeform: (tldr -- a digital approximation of a complex process is 1) not ~the process itself~, noshit.jpg 2) not necessarily all that complex, in the chaos/avalanche sense, or in any way cryptologically hard)
asciilifeform: and if you want reading material, reread the thread where mircea_popescu suggested crypto using transcendental constants etc
asciilifeform: then list things you do ~not~ want
asciilifeform: ben_vulpes: consider the things you actually want from a cryptographic hash
asciilifeform: in a hash for just about any application you want to always live in the former and never, ever in the latter.
asciilifeform: because the transition between 'irreducibly complex' and 'braindamagedly simple' phase space is unknown.
asciilifeform: almost the very definition of terrible hashing function
asciilifeform: ben_vulpes: lulzy contrast of the solid 'hp oscilloscope' goodness of the console buttons, with the crapple turd
davout: asciilifeform: point is piloting a small plane there's just a few things to pay attention to constantly
davout: also "the ball goddammit"
davout: basically "watch your airspeed, watch the fucking airspeed"
ben_vulpes: asciilifeform: i suspect that if i start to explore this risk manifold i will have trouble not ratcheting the risk back
davout: ben_vulpes: piloting when you're actually in the plane is an entirely different thing
davout: which is also why there's yet another minimum speed for a multi engine running on a single engine
ben_vulpes: yeah see when they start talking like this my flugenboner starts to droop
davout: also trinque is right, when losing an engine on a multi-engine you need to apply rudder to compensate for the thrust differential
asciilifeform: (gyroscopic moment of the motor, but also differently-impacting stream from propeller on one wing vs other)
davout: if the wind comes from the right you'll end up landing with right wheel first, then left wheel, then nose wheel
asciilifeform: davout: are there any situations other than crosswind landings/takeoffs where you need yaw-without-roll ?
davout: you apply roll to counter the drift
davout: but that causes the plane to drift
ben_vulpes: note! this does not mean the wings are not horizontal
davout: when landing in a crosswind you basically apply rudder during the approach so that your airplane flies towards the runway, but the nose pointing to the side
asciilifeform: davout: the place that made'em is not far from where i live, it was slowly demolished over decades
asciilifeform: davout: it was a (long-extinct) machine from the age of 'everybody will have an airplane'
trinque: right, not as if you can roll the aircraft when you're about to touch ground (intentionally!)
asciilifeform: ben_vulpes: my current (wholly nonexpert) understanding is that airplane DELIBERATELY omits 'hard interlocks' wherever possible, on the principle that not-being-able-to-X-when-you-must is worse than can-X-when-you-mustn't
davout: maybe i'll understand the answer to this mystery when i become retractable gear certified!
a111: Logged on 2016-11-30 21:04 asciilifeform: i have here a b00k on piloting circa 1940, and already then author insists that rudder pedals are obsolete and have killed a thousand men
ben_vulpes: fuck levers, pressure on the wheels should engage hard interlock
davout: asciilifeform: they *are* very different
asciilifeform: davout: didja ever weigh in on the rudder thread ?
asciilifeform: i'd naively think that this would've been resolved in 1930s, if not earlier, just make the levers vastly different (shapes, or lengths, and feel, etc)
davout: pilot handbook 101: "if after landing you need to apply full throttle to get back to your parking spot, you probably forgot to lower the gear"
asciilifeform: (this is a certainty if the various components do not AT ALL TIMES agree)
asciilifeform: davout, ben_vulpes , et al : it is also tricky to properly rule out the situation where split-trb node behaves like a 'split-brain patient', and external observer gets contradictory answers from it to some possible question
davout: either way, imma head to bed, interdasted in comments re mah wallet cut piece
asciilifeform: they will require a -- quite complicated -- entirely new protocol
asciilifeform: let's take the cuts as specified by mircea_popescu earlier.
asciilifeform: davout: could, theoretically, hurt, if it requires adding 100,000 lines of i/o glue logic
davout: asciilifeform: i think it would be hard to make the argument that a separate binary sitting aside the node could hurt in any way
ben_vulpes: (fwiw i'm down to the last ghostly suggestion, which was to read in the hash as a bignum)
davout: make the miner a separate bin
davout: asciilifeform: i don't thing the argument that the block validation logic can be found in the block validation logic is tenuous
ben_vulpes: unless i misunderstand, the project is truly blocked on making checkpoints configurable.
asciilifeform: gotta be ready to roll on five minutes' notice if the old miners ALL go home SIMULTANEOUSLY.
asciilifeform: now you ~could~ make the -- imho very tenuous -- argument that mining logic is ~implicitly~ present in the block verification logic.
asciilifeform: (a result where there is ~no~ miner available, i exclude from consideration because it pisses on the R in 'trb')
asciilifeform: i historically refused to touch the miner, and will not encourage anyone to touch it, because neither i nor anyone i know is equipped to properly test the result.
davout: but i also get the other point, a lot of that complexity becomes apparent once one actually goes ahead and pops the hood
davout: yet another obvious benefit of amputating the wallet, miner and everything that can pretty obviously done without
asciilifeform: (unfortunately a flow graph where every motherfucking line intersects 55 other lines, is NOT good for anything)
asciilifeform: so that the places to make all plausible cuts become apparentl
asciilifeform: asciilifeform's (and later again jurov's) utter failure to unravel the heap, at least, suggests this.
asciilifeform: davout: thing is, there is no such thing as this 'hot core'
a111: Logged on 2017-01-03 21:29 asciilifeform: though what i pictured is that trb can finally produce the motherfucking ~book~ and it will be possible to start rewrite...
davout: http://btcbase.org/log/2017-01-03#1595836 <<< my image is more like: "trb is this thing from which more and more is removed, until only the radioactive code consisting in ball of tightly packed hot wires which we proceed to put in a little box in which epoxy is poured, and is only interacted with as some black box"
deedbot: http://fr.anco.is/2017/removing-the-wallet-from-trb-first-thoughts << fr.anco.is - Removing the wallet from TRB, first thoughts
asciilifeform: to steal from mircea_popescu's article on subj, 'get inseminated on purpose, rather than 'because hey, there was a party, and i like to drink''
asciilifeform: mircea_popescu: imho it is necessary to 'defile grandfather's pistol' in certain carefully-selected ways, but only if it can be done while making it clear that it was defiled in ~these but not other, unspoken~ ways
asciilifeform: mircea_popescu: these: next!111
mircea_popescu: and what i pictured were 72 cubits high, translucent, ageless, nonmenstruating and deliver pregnancy to term within the day.
asciilifeform: though what i pictured is that trb can finally produce the motherfucking ~book~ and it will be possible to start rewrite...
mircea_popescu: still, the "frozen trb because networking" or "because badlt done block check" etc can't go on forever.
asciilifeform: (it remains to be seen if the thing had ever, or will ever, fit entirely in any head)
asciilifeform: theoretically.
mircea_popescu: still, due to the fact that v allows attribution, the change can be digested over time.
asciilifeform: (if i cannot ~mechanically~ tell that the untouched parts are untouched -- they are, for all intents and purposes, touched)
asciilifeform: the entire point in using a differ in vtron at all (as opposed to signing ENTIRE body of work) is to make the work of the reader tractable.
asciilifeform: this is why the original genesis was such a painful affair (and why it was and remains important to READ how i did it)
mircea_popescu: i suspect graph theory may have a solution for us, but it is not clear to me how.
asciilifeform: mircea_popescu: i'm not entirely certain that the problem is solvable (it IS possible to define a richer diff language that permits block moves, but this also permits inscrutable-to-naked-eye patches to exist.)
mircea_popescu: davout the gordian knot is how to make it both unmagical and self-summarizing.
lobbes: Re: wikipedia. I remember early on in grade school I was actually taught that wikipedia was shit. By the time I entered college this tune had changed. Now it all makes sense
davout: but if it really needs a magical difftron, can it still be said the operator can see everything with naked eye?
mircea_popescu: v is really only as powerful as the underlying differ is.
mircea_popescu: lobbes this incidentally explaisn why wikipedia is such shit - it's ~only function is a sort of open-sourced cliffnotes, and people would much prefer it to be bland and stupidly written so the teacher in class doesn't feel too inclined to think the kids' lifted material isn't his. after all he added all the flavour words in there!