asciilifeform: it occurs to me that one could write a 'garbage compactor' for these
asciilifeform: and they ARE garbage, wtf , 10 decimal coords? thicknesses? WHY
mircea_popescu: exactly. so no, the svg lines aren't garbage - on the contrary, unlike netlist, interesting.
asciilifeform: well theoretically you can make the schem from the netlist. but doing so usefully is np-complete
mircea_popescu: asciilifeform no! they aren't garbage, as you sdaid yourself - nobody would want to see the netlist, because they'd want to see the "unofficial" LINES.
mircea_popescu: asciilifeform which of the two thises are you inquiring about ?
mircea_popescu: a svg also "can not be longer than the total line count"
mircea_popescu: asciilifeform yes but irrespective, it's for the same ~reason~.
asciilifeform: and it has no stinking 10-digit (or any other) pixel coords (wtf), line thicknesses, fonts, etc.
asciilifeform: mircea_popescu: think, netlist is the essential 'soul' of schematic, definitionally such that if you change ONE character, you get a ~different~ (and , likely, broken) device.
a111: Logged on 2016-12-12 13:38 mircea_popescu: how the fuck is it that those supposedly ubervaluable engineers that don't exist anywhere else and can't be replicated (as per microsoft idiot) nevertheless DO NOT COMPREHEND the difference between the present conditional and the pluperfect conditional.
asciilifeform: in our case, it is shorter than the pld src
a111: Logged on 2016-12-12 13:10 mircea_popescu: also the reason your handmade netlist is not usable is ~same svg is not usable : too long.
asciilifeform: http://btcbase.org/log/2016-12-12#1581961 << this is pointedly not so, think about it for a minute, a netlist cannot be longer than the active pincount !
mircea_popescu: fuck 'em. they want to get into engineering school, first they acquire literacy in their mother tongue what the everliving fuck might this have had being!
mircea_popescu: then they choose to write in brainfuck and everybody goes "oh, brainfuck, modern language" rather than "oh, these idiots can't even speak english."
mircea_popescu: it may be, it might have been, whatever the fuck, same thing.
mircea_popescu: how the fuck is it that those supposedly ubervaluable engineers that don't exist anywhere else and can't be replicated (as per microsoft idiot) nevertheless DO NOT COMPREHEND the difference between the present conditional and the pluperfect conditional.
Framedragger: yeah, would be great to have ssl certs etc all in the same place, timestamped, so one could track history, to an extent. (and then be able to offer realtime scans and alerts as a service, say...)
mircea_popescu: this is becoming quite the autonomous "net awareness" item.
Framedragger: i guess the scanned banners could go into a separate "scans" table with timestamps / scan event numbers, so that one could JOIN and check multiple banners for ip, especially when i plan to later re-scan everything again, etc.
Framedragger: the later "rescan" only added previously-unseen IPs, + new IPs. same ip was never scanned twice, assuming it already spat out a banner and pubkey the first time.
mircea_popescu: ah. because boxes with the various "autoupdate" things mya do that.
Framedragger: btw i'm going thru those ssh banners from ssh scan logs finally, and there's some inconsistent crap there (thanks openssh): same ip&port may respond with two different banners during same scan (the ssh-keyscan utility may spit banners for same server multiple times). it seems usually the mismatch is in adding a minor version onto ssh server string only (e.g. [SSH-2.0-OpenSSH_5.8] vs. older [SSH-2.0-OpenSSH_5.8p2])
Framedragger: (thanks for the heads up)
mircea_popescu: also the reason your handmade netlist is not usable is ~same svg is not usable : too long.
Framedragger: (a rather decent bitcoin invoicing interface btw)
a111: Logged on 2016-12-12 03:46 phf: fwiw that solution works reasonably with btcbase (right now it's not used, but it picks up a README from each patches folder and cats it inline before the graph, so i can add a "here's all the additional stuff that you might need for this v tree")
ben_vulpes: http://btcbase.org/log/2016-12-12#1581929 << http://btcbase.org/patches?patchset=bot not to make more work for you, but the smooshing of ircbot and logbot apparently hosed your pre-graph catment
asciilifeform: i considered making, by hand, a v-able netlist, with component values etc. , in sexpr format, for a nonexistent tmsr cad. the real problem is that everyone, without exception and including myself, will insist on viewing the 'unofficial' pictoral schematic.
mircea_popescu: i fear the only way forward is for data to be base64'd, deedbotted, and then referenced as such in code.
mircea_popescu: including data with the code is a nightmare not least because it puts state into things
a111: Logged on 2016-12-12 03:49 phf: you verify vpatch using its sig, you verify additional assets using their sig. you can't verify that the set is exhaustive though
mircea_popescu: yeah ; mp-wp had no themes did it.
phf: fwiw it's the same with wordpress
mircea_popescu: (i've been running into a version fo this in eulora too ; with the art assets)
mircea_popescu: bundling data with code is a serious problem ; the v model brings it in a very sharp focus
phf: you verify vpatch using its sig, you verify additional assets using their sig. you can't verify that the set is exhaustive though
phf: fwiw that solution works reasonably with btcbase (right now it's not used, but it picks up a README from each patches folder and cats it inline before the graph, so i can add a "here's all the additional stuff that you might need for this v tree")
mod6: That said, I'm not positive what is a favorable solution to this. For me, I guess I would have considered a disjointed genesis. All code in a genesis.vpatch, plus a comment in the code or README.txt file that points to a clearsigned, base64 encoded deed of the (repeatably extractabale) binary (image in this case).
mod6: All in all, I agree that blobs do not belong in a vpatch. As stated, they are for readable, grokable, text only.
mircea_popescu: http://btcbase.org/log/2016-12-12#1581907 << which is the very fucking point, of course. "all equal, all stupid, all vorvarts!"
a111: Logged on 2016-12-12 02:52 asciilifeform: he didn't release the raw slides ?
mircea_popescu: http://btcbase.org/log/2016-12-12#1581905 << apparently knuth is blisfully unaware the micros even exist.
a111: Logged on 2016-12-12 02:51 phf: i will flatten this shit, or help me god. so far the solution i figured that doesn't require writing code or using dodgy third party software is to use the video as a material texture inside a sphere in blender. with some 3d space camera shuffling i can produce two separate video streams, one of knuth and one of slides, but the result looks like dog so far.
asciilifeform: he didn't release the raw slides ?
phf: i will flatten this shit, or help me god. so far the solution i figured that doesn't require writing code or using dodgy third party software is to use the video as a material texture inside a sphere in blender. with some 3d space camera shuffling i can produce two separate video streams, one of knuth and one of slides, but the result looks like dog so far.
phf: of course the only proper way of watching this video is using Google Cardboard! https://vr.google.com/cardboard/
asciilifeform: 'By using a computer mouse, or by tilting a smartphone or tablet, participants will be able to individually control the video angle and experience 360-degree viewing.'
asciilifeform: now if only there were also a way to pump the unwashed student's armpit flavour into my office also !!
asciilifeform: '“This year, our Learning Innovation team is going to be piloting the use of a new virtual reality 360-degree video camera that will actually allow all those people tuning in over the internet to attend the lecture as if they were physically in the auditorium,”'
phf: in some usable format (short of installing google chrome of course) and so far "transcoding with a deskew shader" seems like the most viable option ???
phf: so stanford release knuth's annual lecture in a "360" video format, and the write up is disgusting "stanford part of the future!!1". basically you get a 7gib video/80mib audio stream, a fraction of which is a distorted slides and knuth, where's the bulk of which is audience and static ceiling. the thing doesn't render on anything but google stack and ipads. i spent about two days jumping through various hoops trying to get it to render
jhvh1: shinohai: The operation succeeded.
mircea_popescu: to justify why there's no legal recourse for bad software - and guess what, it stands just as well to explain why it is better to murder and rape than to microsoft.
mircea_popescu: ~maybe~ the girl really like it ; and ~maybe~ the dude had it coming to him. but ~certainly~ there is no room for windows & co in this world. the advantage of software, it allows definitive statements to be made like that (an inverse of the problem of liability, wherein hardware lock making company CAN replace any broken locks it shipped, but software lock making company can't because they're all the same lock.) this was used
mircea_popescu: because well... he is. and if you take the ethical stand that you don't want to be paid by rapists and murderers, then microsoft tops the list in any reasonable interpretation and that's what it is.
mircea_popescu: so in short : no, it's not ok to take microsoft money. human trafficking money, the proceeds of sales of endangered baby panda tears, the ransom paid for the return of obama's kidnapped daughters, all that is a-ok.
mircea_popescu: im sure koch got >1k and IM FUCKING SURE microsoft is worse than any "anti-gay-marriage racists" out there.
mircea_popescu: Framedragger hey, do you also bitch at people who bitch at, eg, mozilla guy who happens to have invented the shit mozilla runs being "bad" because he donated 1k to whoever ?
asciilifeform: if my sig is of 'A35A727799941F46F4500F25389F21F4E995F64AC65341080052EC014A8BACD76D992D0C7A5B0250502D', that thing will appear ~to naked eye~ in the pcode as 'zA35A727799941F46F4500F25389F21F4E995F64AC65341080052EC014A8BACD76D992D0C7A5B0250502D' .
asciilifeform: (pcode can output, if it weren't obvious, another pcode)
asciilifeform: e.g., want to sign hash? get it whereever you like, and drop it IN THE PCODE
asciilifeform: it is POSSIBLE to write, e.g., sha512, or keccak, in p, but not essentially necessary, because one of the philosophical turtles of p is that you SEE what you are signing
asciilifeform: there are no special-purpose ops to produce hashes
asciilifeform: i was not going to expand on the 'p' thread until the proggy is done, but this is probably a good time to say 1 more
asciilifeform: when they are poor; but 100x easier when poor+stupid also.
a111: Logged on 2016-12-11 21:50 mircea_popescu: and then koch wants to go around pretending like he's one of us.
asciilifeform: http://btcbase.org/log/2016-12-11#1581840 << i dunno that he even bothers to pretend, he is a confessed recipient of microshit cheques
a111: Logged on 2016-12-11 21:50 trinque: and the process is not reversible, couldn't even put the information in like "this had lf" somewhere
trinque: and the process is not reversible, couldn't even put the information in like "this had lf" somewhere
mircea_popescu: and then koch wants to go around pretending like he's one of us.
trinque: clearsig "canonicalization" includes line ending conversion to crlf before sig, and lf-only endings probably represent *all* submissions to the ML
mircea_popescu: trinque at the very least a result here could be where you machine-verify that X item deedbotted is equal to X' v-item through the process of a) X valid clearsigned b) X' valid detach0igned c) X = preprocessor(X')
mircea_popescu: trinque the clearsign iirc is a stricter operation ; but plenty of signedtxts emerge untouched from both processes.
trinque: so then if I've got a sig of an un-transformed item in ML, it may contain things the transformer would've stripped/modified if it were clearsigned
trinque: mircea_popescu: it seems to me the two paths (clearsign and detached-sign) sign completely different piles of bits, if the clearsign first goes through some transformer
mircea_popescu: oic. well it's the first here.
asciilifeform: mircea_popescu: this is neither 1st nor 10th 'faux computer' i make.
asciilifeform: you ~compose~ p operations by physically concatting the p-progs.
trinque: and whatever else it does to the poor clearsigs
trinque: so then the original item would have to have been detach-signed with line endings already mangled to get a clearsigned version that'd verify
asciilifeform: to verify a sig, you feed ptron this VALID PROGRAM ^ -- which creates the pubkey in memory -- and then on top of that, the sigogram
asciilifeform: what this does is store the modulus in variable #13 ('m' simply pushes integer 13 to stack, '!' stores top of stack in var # second-of-stack)
trinque: I'd have to spelunk further to confirm, but it looks like clearsigning is canonicalize -> sign
mircea_popescu: ("find the longest ascii file so that a) lzw compression is under 1% and b) the encryption scheme therein discussed is trivially breakable)
asciilifeform: for instance, p code is required to first thing put on the stack 1) how many p-instrs are permitted 2) how many bits wide the fixints are
asciilifeform: there are things that are not in p yet but have to be
asciilifeform: 0x20 and below rather
trinque: mircea_popescu: reading up on the mangling process, maybe
phf: wait so X multiplies da9 and 100, = compares the result with 2ff, 9ae???
asciilifeform: phf: http://wotpaste.cascadianhacker.com/pastes/ApmIh/?raw=true << same example but before i stripped out the comments
mircea_popescu: ah yes, there is that. so can you try an actual messagelist item ?
asciilifeform: phf: it is the known result
trinque: mircea_popescu: did, the signed item has to have already been run through gpg's idiot preprocessor
mircea_popescu: trinque theoretically at least it should be just transform the binary sig into a plaintext packet and glue it in.
asciilifeform: to make the parser almost nonexistently simple
mircea_popescu: trinque care to look into if you can bash the signature out of detachsig and attach it as clearsigned ?
asciilifeform: = takes 2 numbers off stack and puts 0 if they are equal, etc
asciilifeform: thereby you get a stateless hexbignumreader
asciilifeform: 'z' simply puts a 0 on the stack, and if ptron sees 0-9 or A-F it shoves the 0 leftward 4 bits, and stuffs in the nibble in place
trinque: new features could come on the P side.
asciilifeform: afaik: not with standard gpg, or such that standard gpg will eat the result
asciilifeform: i dunno if there exists upon this cursed earth a rope that doesn't break when i go and pull on it...
mircea_popescu: it just eats the patch straight neh ?
asciilifeform: is there any means for doing this ?
mircea_popescu: "cold equations" is bad enough when you ~fuck up~, but let me tell you about when you didn't even fuck up in the first place.
asciilifeform: the schem will be nonclearsigned, the sig -- posted. the boards -- physically wrapped in copies of the schem. is all i can offer.
mircea_popescu: it'd be too fucking easy a hole for adlais to slither into.
mircea_popescu: hey, i don't want unreadable turds in v trees either.
asciilifeform: and tmsr-fab, and many other things.
mircea_popescu: yeah. but do you agree the actual problem is the lack of a tmsr-cad ? as far as "right things" go ?
asciilifeform: simply that i am stuck in this corner, where there is a thing that needs doing, that i do not have with what to do.
mircea_popescu: but if that's what it is - neither v nor deedbot need fixing for it.
mircea_popescu: so then tmsr-cad ?
asciilifeform: the actual semantic content of that schem is miniscule.
mircea_popescu: give me something other than a 5mb dxf to work with, as per this "Short when i wrote it" ?
asciilifeform: then what.
jurov: it's just an attribute. when i include some TEXT chunk in png with an url, nothing happens either.
mircea_popescu: holy shit you're going to argue from "theoretical logic" ?
jurov: asciilifeform: the mircea's file does not have dtd declaration anyway...
mircea_popescu: so no, png in no sense better than svg in this theoretical approach.
mircea_popescu: yes, per image load - if you don't have them, image won't load.
asciilifeform: i do not want to sign a 'insert usg turd here!' even if theoretically the instruction is not followed on extant renderer
phf: asciilifeform: nothing at this point, none of the xml parses uses DTDs (as predicted by naggum)
asciilifeform: libpng doesn't ask for a turd from the www afaik !
asciilifeform: the problem of signing shit soup.
asciilifeform: possibly, but does this solve the problem.
mircea_popescu: asciilifeform i could readily reduce the svg if you wish, most of it is dreamweaver-style pointless tagging
asciilifeform: which i am not even sure differs substantially from the svg situation.
asciilifeform: the original, btw, is 600+kB of xml. but this does 0 without the closed-source turd viewer.
asciilifeform: and ftr i am not signing a document that i am supposed to be the original author of but CANNOT GET TO HAPPEN ON MY BOX
mircea_popescu: ill report in a sec, machine currently hosed painting the svg over 200k x 200k canvas
phf: mircea_popescu: i have nfi either, i have it unpacked
mircea_popescu: you first debased then concat'd and it still worked ?!
asciilifeform: the second --- not
asciilifeform: btw the first svg is cut off mid-tag
jurov: i de-base64 each one, then concat and i can open it fine
mircea_popescu: jurov did you concat them correctly ?
mircea_popescu: well the dxf you produced - is
mircea_popescu: there's no way to make an arbitrarily complex schematic fit in an arbitrarily short space.
asciilifeform: fwiw inkscape won't eat it either
mircea_popescu: asciilifeform look, saying "the v is not allowing x char sequence i wanna use" is one thing. saying "o woes i have this here book of 5mn chars and i want it to be readable in 3 minutes" is another thing.
trinque: alrighty then
asciilifeform: i have '---' in there. it will FUCKING STAY '---'.
trinque: I read the whole log.
mircea_popescu: it seriously looks good. even your 10K for R5 is there
asciilifeform: (others -- work)
trinque just made it to the surface.
asciilifeform: what the hell didja read'em in
asciilifeform: btw these render nowhere mircea_popescu
asciilifeform: the unreadable dxf yes
mircea_popescu: http://wotpaste.cascadianhacker.com/pastes/PfVG0/?raw=true + http://wotpaste.cascadianhacker.com/pastes/1Pq8P/?raw=true then de-base64.
asciilifeform: http://nosuchlabs.com/pub/tmp/goat.dxf << and i will not that i HAVE NFI WHAT THIS LOOKS LIKE or if it corresponds to what i saw when i drew the fucker!!!!
asciilifeform: cheap but then folks would have less incentive to use the turdware and That Would Be Wrong
asciilifeform: just as there are no tmsr chip fabs, etc
a111: Logged on 2016-12-11 19:14 phf: binary blobs remain an unsolved problem. the whole idea of binary (or base64) is contrary to some of the more philosophical aspects of vpatch you and mp like to discuss, so i don't know if base64 even solves it.
mircea_popescu: see, there's many problems here, and phf correctly identifies the general case in http://btcbase.org/log/2016-12-11#1581294 ; for eg : what if i want to scale ?
asciilifeform: also dxf apparently, but i have not been able to get ~these~ to render anywhere!
asciilifeform: take it up with the german schmucks
asciilifeform: they simply aren't
mircea_popescu: asciilifeform why aren't these schematics svg anyway btw ?
mircea_popescu: i was wtf is this extra shit on the side oh OH!
asciilifeform: and haha, there are TWO +++ lines !
mircea_popescu: just as an artefact of the lifes and times.
asciilifeform: i will illustrate: yes, cad proggy can export netlist. but said netlist has no provisions for describing a) part valus (e.g., you get C10, and no place for '0.47uF') or the geometries of the physical chips (you get, e.g., IO1-15 but way to say 'that's pin 44 of a pqfp-44 plastic thinleadc carrier)
mircea_popescu: first fuckgoats, then laundrygoats!
mircea_popescu: while xml is shit, it's not directly clear that the correct move here isn't, eg, to make a tmsr-cad format.
phf: so my proposed "put everything into prelude" solution preserves the literacy angle, and uses prelude as a "annotation" section, i.e. space for verbiage, supporting material etc. this will work for free with things like btcbase. can also pack it into a tar file and sign ~that~, but breaks a lot of tools
mircea_popescu: ie, the reason koch-gpg-clearsing worked ok for us for a long time is that while flawed as alf correctly (and repeatedly for a year now) points out, nevertheless its hole falls atop a hole of v, namely that it doesn't do "-----"
mircea_popescu: yes, they are.
phf: mircea_popescu: well, the way i got what you grokked from v when it originally came out, is that vpatches are literate, for things that you can reason ~and meaningfully amend~. "unicode on a chalkboard" thread is also relevant
mircea_popescu: the whole point we're trying to get out of is that adhocness
mircea_popescu: so then "it always can be done in particular" is no asnwer, is it ?
jurov: there's also svg using data uri, where you can put all data on one line, thus avoiding +++ on line start problem
asciilifeform: because it hides the problem.
asciilifeform: jurov: i could trivially convert it to a shell script that 1) vdiff will eat without incident and 2) produces the original png when run. but this is morally problematic.
jurov: i don't see what's the problem with having 8-10x more data
phf: mircea_popescu: i get the problem, but you can always some up with a strict no-blob solution, whether or not it's practical. in this case include your entire training set, that, being go boards, can literally be represented as ascii x/o drawings
mircea_popescu: asciilifeform incidentally it always amuses me how fascinated the redditard/ycombinatard always seem to be with the "pick 2 of 3" sort of shits, to the point of aspiring to have one named after themselves / renaming ancient ones in anecdotal rephrasings to honor whatever particularly hipstery doofus. never once does it occur to them that the only thing this means is "the problem's misstated".
mircea_popescu: phf suppose you make an ai expert system to beat us at go. this gives you two practical options : either include 10gb worth of binary flags preset ; or else have us beat it at go for 10 centuries before it gets to where it plays like a freshly fucked 19yo.
asciilifeform: phf: there is only 1 problem , it leaves me with a useless genesis.
phf: i'm convinced that blobs shouldn't go into vpatches, this completely breaks the whole readability/communication aspect of it, while still remaining promisestronic: you have to take some additional step after pressing to get the final images.
mircea_popescu: jurov no because then you're stuck with a reflexive system
jurov: MIME avoids this by generating unique magic strings as delimiters after the fact. but you don't want to, i guess.
mircea_popescu: so i'ma just wait here for you to p, and comment thar rather than try and spec a tmsr-rsa-clearsign ?
mircea_popescu: (i confess it does to me, has no value other than cheapness. but gotta see what peoples think too!)
asciilifeform: *therein
asciilifeform: well yes. any magic string can make the all-time-log undeedable.
asciilifeform: (how? someone will sign the log! and here it is, mircea_popescu wrote it!)
mircea_popescu: in principle we should just make it the whole story, "----- BEGIN FUCKAGE -----"
mircea_popescu: incidentally asciilifeform : as a palleative : could we actually just fucking edit koch-rsa so that "----" rather than "-" becomes a symbol and deploy this on deedbot ?
mircea_popescu: you don't "exponent multiply" the payload "up until you run into the number of the beast"
asciilifeform: i, with own two hands, will kill them.
asciilifeform: they WILL die.
mircea_popescu: phf not so. the rsa part is NOT magic char driven.
phf: mircea_popescu: the whole thing is "magic char" driven though. in this particular case the failure is not just from clearsign, but from a combination of clearsign and vpatch own magic chars.
mircea_popescu: (hey, it's not THE most common english letter!)
mircea_popescu: this may even be ok if the magic chars weren't fucking DASHES ffs.
asciilifeform: mircea_popescu: and wait till you get in the log to the OTHER inbandism idiocy of the day
mircea_popescu: phf these are palleative solutions, you understand i'm sure. the fundamental problem here is that clearisgn is a "magic char" driven thing like nonsense "nullterm" string constructions
phf: so vpatches all have preludes, by virtue of how diff/patch works (that's how you can just cat mail.mbox > patch). i was thinking of using that prelude for readme, but you can put base64 binary files there, and verbally communicate additional required steps. it's ugly, but it's without mutilating core concept. prelude is reveserved for whatever ugly special case hacks, etc.
mircea_popescu: would you agree the correct move here is to actually specify clearsign format for tmsr-rsa ?
asciilifeform: do they even v ?
mircea_popescu: http://btcbase.org/log/2016-12-11#1581240 << the never part there is iffy. i ~think~ that once we finally get tmsr-rsa, we will NOT be doing that idiocy.
asciilifeform: and how do the eulora folks solve this..?
phf: asciilifeform: i've got another one :)
mircea_popescu: adlai your attempts to state "the moon math" for $item resulted in knowledge that... "not possible to put into bitcoin" ? this is mixnonsense wtf. math is math.
adlai: (this is re: the "MimbleWimble" not-yet-altcoin)
adlai: re: "go do things", another thing i've been doing is studying towards http://btcbase.org/log/2016-10-19#1556804. my understanding thus far is that it is firmly within "not possible to put into Bitcoin" territory, but still potentially interesting; and i am not yet confident enough in my understanding to try explaining to others; but i should be in a ~month.
asciilifeform: but fg-genesis is an item i produced out of thin aether. it is v material.
a111: Logged on 2016-12-11 18:40 asciilifeform: WITHOUT THE IDIOT MUTILATION of clearsign
asciilifeform: it is the worst possible answer.
asciilifeform: phf: mutilating and complicating vdiff or v is the very last thing i'm interested in.
mircea_popescu: just drop the whole imagination thing, it ain't useful. it just provides a way for you to waste your energy. go do things, if you ask then "hey, i am now opening shop for adlai's electronic dildos, but i don't intend to do X like nsa did, but this other way, anyone see a problem?" you'll get an answer and proceed from there.
asciilifeform: i'ma ask mircea_popescu to make the cut.
adlai: mircea_popescu: i'm trying to replace the mistaken assumptions, stemming from operating hypothetical services within my own imagination.
asciilifeform: phf: i could easily write proggy that eats a binturd and produces a vdiff-harmless shell script that, when executed, disgorges same. but the IDEA, yes, bothers me
deedbot: shinohai updated rating of BingoBoingo from 2 to 3 << "Qntra: Herald of The Most Serene Republic"
phf: asciilifeform: perhaps binary blobs don't belong ~in~ vpatch in general for philosophical and mechanical reasons. my proposed solution 2 would be to extend vpatch tool to tack one base64 binary blobs after the fact. you can grep for "appears to be binary" or whatever it is and feed those separately to uuencode/decode.
mircea_popescu: your assumption is mistaken in your having one in the first place.