log☇︎
2500+ entries in 0.253s
asciilifeform: as for the multitude of vtrons, there is absolutely nothing wrong with it, but i am seeing folks suffer from easily-curable issues that were not present in my original prototype (which, recall, had NO state other than patch/seal/keys working set, and did not use the net, and had the 'origin' command, etc.)
asciilifeform: it is actually quite feasible to have a patch that 'touches everything' but is visually inspectable in three minutes - e.g., mod6's proposed 'release patch' that merely tacks a comment to the top of every file.
assbot: Logged on 31-01-2016 14:48:21; mircea_popescu: http://www.xenosystems.net/chaos-patch-99/ << this guy is fucking epic lol. "go is gone. Yudkowsky sounds the alarm." and links a fb posting.
assbot: Outside in - Involvements with reality » Blog Archive » Chaos Patch (#99) ... ( http://bit.ly/1KPtcPz )
mircea_popescu: http://www.xenosystems.net/chaos-patch-99/ << this guy is fucking epic lol. "go is gone. Yudkowsky sounds the alarm." and links a fb posting. ☟︎
mircea_popescu: just, if you want to add a patch, should be able to dump it as dpaste also. ☟︎
ben_vulpes: a full sync of each bitcoind produced by each patch sent to the ml
ben_vulpes: i shut the jenkins down because full syncs of each patch was prohibitively resource and time intensive and parallelized poorly
punkman: http://log.bitcoin-assets.com/?date=30-01-2016#1390093 would provide hands-off testing to see whether your new patch compiles to different archs, with different libc, etc. think jenkins ☝︎
mod6: and... its maybe slightly different since the first patch was inclued in the first release.
mod6: and i'll also have to "re-grind" my high/low S patch
mod6: <+pete_dushenski> ver 99999 is fine by me << yeah, this is fine. but the fact that it doesn't do what it should do isn't. i'll pull the patch tomorrow.
asciilifeform: likewise it appears that the shiva patch has an issue, you end up having to LC_ALL="C" ./bitcoind -shivainit=./shiva/init.scm -shiva getinfo
asciilifeform: will be happy to produce an antimatter patch for it if this is needed.
asciilifeform: mod6: i officially pronounce the verstring patch to be buggy and unfit for inclusion in its present state
phf: "patch" is surprisingly trivial once you get past unified diff format parsing, http://paste.lisp.org/display/306234
ascii_rear: but, 1) there is a skull on the patch for a reason 2) there is not actually an alternative to tinyscheme.
phf: i pressed it programmatically, without "patch" utility, before leaving home, so didnt have time to test it
phf: ben_vulpes: so actually inside the patch hunks are grouped by files, what i was referring to as hunks are actually per file hunk groups. so in other word the graph is already edge-per-file. i added the labels but it looks dodgy http://glyf.org/tmp/labels1.png http://glyf.org/tmp/labels2.png
ben_vulpes: first: node.js. second: "naw, vpatches are for faggots". third: 2.5kloc patch degendering reference implementation log statements.
phf: and the one ascii requested, transition graph for when there's only one hunk per patch file http://glyf.org/tmp/trb-transition-hunks-asplode.png (i filtered out all the genesis files that don't have further transitions)
shinohai: I know the patch applied, because bitcoind -? shows it as an option
assbot: Logged on 21-01-2016 15:48:48; asciilifeform: mod6, shinohai: the version string patch works. BUT you will always see the default version in the boot log!! because said printf ~precedes~ the setting of the custom string. this may explain shinohai's confusion.
pete_dushenski: mod6: i have a question about the version string patch... i seem to have applied it correctly but i'm still seeing version "50400" even when trying the '-setvernum' command so i'm not sure what i'm doing wrong. this is a new instance that's in the process of syncing, not 'tevye', which properly shows '99999'. i used the original 99997k script and applied 'version-strings' and 'hearnificarum' manually and both are now
mod6: # Required Public Keys to add to your PGP Keyring. These are used to verify V, rotor, buildroot, and trinque's patch:
mod6: # Required Public Keys to add to your PGP Keyring. This are used to verify V, rotor, buildroot, and trinque's patch:
phf: having multiple hunks in a patch modify several files at the same time creates a harder transition constraint then simple antecedent/descendent pairing, so i'm thinking it's probably safe to construct a graph in terms of transitions rather then purely a/d. the end result should be the same
mod6: kako: http://therealbitcoin.org/ml/btc-dev/attachments/20150808/rotor-db-configure-fix_a955ba9174ccb17790dc9d7c1e2a61794a1c803d.patch
phf: mircea_popescu: anyway, i was speaking of how things are right now in v.py/v.pl. a hunk touches a file, hunks are grouped thematically (i.e. by a single unit of meaning) in a patch. a pairing of antecedent/descendent creates a graph and in combination with topo logical sort produces a press order
kakobrekla: what is the 'special patch'?
mircea_popescu: + # You need to have these keys in your gpg keyring to vertify V, trinque's special patch, buildroot, and other stuff.
mod6: to verify V, trinque's special patch, buildroot, and other stuff.
phf: i remember ascii suggesting that perhaps we should stick to one hunk per patch, but that's not the case right with existing patches. if z thematically needs to touch two different files (a and b), then both modifications are inside a single patch
ascii_butugychag: mircea_popescu: note how i cut the shiva patch in 2
mircea_popescu: the best way forward, imo, is to restructure it as 1) one large patch consisting of nothing but the return fixes and some smaller patches doing the rest of the stuff built on top of that.
mircea_popescu: polarbeard don't get this wrong, you did a lot of very useful work in that patch.
mircea_popescu: note that the only way i got my apparently very controversial /s /t patch to even be considered was showing how it could be machined.
ascii_butugychag: so i can ask the same question, line after line, for a whole patch
mircea_popescu: polarbeard yeah, you get the right idea, but seriously, making a single patch consisting of JUST -return error +return error is the right way
mircea_popescu: but it's also immense and should all be one single patch so we can see it more readily.
polarbeard: mircea_popescu: the patch ended up being a bit verbose, yep, but it made sense to me fixing the messages as I was adding metadata, that's why I didn't add filtering, to not add actual logic to an already pretty big patch
mircea_popescu: there's a lot of stuff in there that will necessarily result in a huge patch (the guy is stuck, willy-nilly, doing a lot of
punkman: mircea_popescu: so how would you split polarbeard's patch?
ascii_butugychag: pete_dushenski: supposedly there is an x11 patch to make it go
mircea_popescu: so in full terms, i would say that again, including both in the same patch is both premature optimization and a kludge.
phf: a way to indicate (a->a' b->b') would normally be to include them both in a single patch. i think that's how it's done now
phf: so in the above example x y z are distinct patches, their effect is to take files through states, a, a', a". so if a patch takes a->a', then it needs a in genesis. if a patch takes a'->a" then it needs genesis and a->a' patch. i was trying to indicate that y and z modify two separate files, i'm trying to see what is there that indicates that state (a' b') is desirable over (a' b)
punkman: phf, my "shortest path" algo is to recursively find all antecedents of the patch I want to press. it's useful.
ascii_butugychag: phf: i have a patch for adding actual tcp to ts
ascii_butugychag: so, e.g., mircea_popescu's retabbing, would be a five-line patch
mod6: as far as comments in the vpatch that are mearly just for archaeological or annotative purposes, i think something could be done here especially when we get our own diff/patch.
ascii_butugychag: shave half of face, submit patch, shave other half, etc
mircea_popescu: now, one obvious solution (wasn't obvious to me at the time what it was a solution to) is to INCLUDE COMMENTS in the patches, like i did with my only signed patch i didn't really intended as part of a final pressing
mircea_popescu: approach thematter from the other, practical side, maybe it's easier that way. so i made a patch that did X, then i made a patch that did X'. i consider X' to be > X.
ascii_butugychag: but i will note that introducing a new patch that is binary-identical to an old one, is an error.
ascii_butugychag: so if i submit frobs-x.vpatch and also-frobs-x.vpatch and they both reference patch frobbed-x-before.vpatch, then i can press to frob-x OR to also-frobs-x or to any other leaf.
assbot: Logged on 29-01-2016 10:29:19; jurov: I know that i should manage the repository myself, but imo a version field would be good to have so that I can easily spot if someone posts revised version of the same patch.
ascii_butugychag: mircea_popescu: i closed the gap today, see the next patch.
assbot: Logged on 29-01-2016 10:02:07; punkman: so shiva patch wholly includes the tinyscheme genesis instead of referencing it?
ascii_butugychag read the mega-patch-of-all-ages yet ?
assbot: Logged on 29-01-2016 07:42:24; polarbeard: I've timestamped and categorized trb log lines, as well as improved the messages and removed (seeming never activated) destructive log rotation, here is the signed patch: https://gist.github.com/polarbeard/961aa315c69b89d2e613 in case somebody with a decent reputation wants to review it so I can submit it to the ml
ascii_butugychag: (as my instructions note, it is necessary to rename a directory after pressing tinyscheme-genesis to end up with the shiva subdir in shiva patch)
assbot: Logged on 29-01-2016 10:02:07; punkman: so shiva patch wholly includes the tinyscheme genesis instead of referencing it?
asciilifeform: ;;later tell polarbeard when writing a patch, consider the effort required to properly grasp EVERY LINE
assbot: Logged on 29-01-2016 08:08:31; ben_vulpes: "this 2.6kloc patch rewrites every log line in the reference implementation, in support of further work i'm doing on" etc etc
assbot: Logged on 29-01-2016 07:42:24; polarbeard: I've timestamped and categorized trb log lines, as well as improved the messages and removed (seeming never activated) destructive log rotation, here is the signed patch: https://gist.github.com/polarbeard/961aa315c69b89d2e613 in case somebody with a decent reputation wants to review it so I can submit it to the ml
punkman: I think it'd be best to name the 2nd patch mempool_frobsv2.vpatch or something like that
jurov: and the version field as just advisory indication which is the more recent patch
polarbeard: ftr, I've updated the gist, will send the patch to the ml later
jurov: I know that i should manage the repository myself, but imo a version field would be good to have so that I can easily spot if someone posts revised version of the same patch. ☟︎
punkman: so shiva patch wholly includes the tinyscheme genesis instead of referencing it? ☟︎☟︎
assbot: Logged on 29-01-2016 08:13:49; ben_vulpes: definitely split the actual muntzing out into its own patch instead of complecting it with the rest of the work. i might be able to read and meditate on that patch.
ben_vulpes: removing dead code is good, far better than rewriting the log lines. just split it out so that it can be reviewed seperately from the two thousand five hundred and ninety odd other lines of that patch.
ben_vulpes: definitely split the actual muntzing out into its own patch instead of complecting it with the rest of the work. i might be able to read and meditate on that patch. ☟︎
ben_vulpes: "this 2.6kloc patch rewrites every log line in the reference implementation, in support of further work i'm doing on" etc etc ☟︎
ben_vulpes: you realize this is a 3kloc patch.
ben_vulpes: the problem with a patch of this size is that my face wearies of reading myriad log lines changed for seemingly aesthetic reasons, and i fear that even a close read will miss an underhanded c contestant.
polarbeard: it truly is, as what value it has, it's just a step into better debugging capabilities, for this patch I've not added new log lines on purpose but I plan doing it for the next
polarbeard: I've timestamped and categorized trb log lines, as well as improved the messages and removed (seeming never activated) destructive log rotation, here is the signed patch: https://gist.github.com/polarbeard/961aa315c69b89d2e613 in case somebody with a decent reputation wants to review it so I can submit it to the ml ☟︎☟︎
asciilifeform: (and yes i will re-grind the ts genesis later. it needs to be re-based to add up to patch 1 above, also)
mod6: ok you should probably wait to run vdiff against the changes until i've got the release patch out then, so you can start from there.
mod6: im thinking that i'm gonna tag the release, and publish the release patch soon. just hoping that if any of it touches trb, that it'll be post-release patch so there are no conflicts.
thestringpuller: once I'm up to sync, i'll patch that in and restart the node
thestringpuller: is the version string patch going into the newest release?
assbot: Logged on 27-01-2016 21:38:09; mats: i was just in a meeting where bosses decided to downgrade ~10k windows 10 users to TLS 1.0 (from 1.2) because of a bug from patch tuesday
mats: i was just in a meeting where bosses decided to downgrade ~10k windows 10 users to TLS 1.0 (from 1.2) because of a bug from patch tuesday ☟︎
jurov: no, from my own mempool patch
ascii_butugychag: jurov: this from mod6's patch?!
shinohai: node is now running mod6 's latest patch, running beautifully \o/
mircea_popescu: look in the ml for the patch for that.
assbot: Logged on 25-01-2016 01:30:26; *: asciilifeform observes that this thread has already taken up more space than the patch.
mod6: alright, the DER patch has been sent to the ML:
mod6: hmm. so then, I believe if i remove the tabs from my S-patch and leave the spaces ~in~, then it'll look more like Mr. P.'s patch he put together.
asciilifeform: polarbeard: didja write a patch and i missed it reading the log ?
mircea_popescu: oh ok, my bad. so what i was thinking here is : a) finish the patch ; b) do all the other stuff you were going to do in your own time ; c) when ready make a new version ; d) take a week off, during which everyone [who cares to] can write and run a script to / /
mod6: But it probably would have been the last thing to go into the tree before a release patch.
mircea_popescu: i guess i unwittingly confused you, sorry about that. the S-patch was not actually intended to mark a new release did it ?
mircea_popescu: what's your current work on teh patch ?
mod6: Ok sounds good mircea_popescu. I'll stop my current work on the high/low patch. I'll spend this week getting the changes [.wot in the pwd, and mechanical post-patch hash checking] in V [v99996] released. Then I'll immediately begin work on a re-alignment of all the patches we currently are distributing.
ben_vulpes: perhaps fixing magic numbers is out of scope for a patch to force s-values?