PeterL[billymg]: hello, I have upgraded to the latest blatta-9985, seems to start up just fine
awt[billymg]: PeterL: nice!
awt[billymg]: Seems like dupes are dead.
awt[billymg]: 9984 and 9983 are out if anyone wants to try. Fixes /AT and adds /KNOB.
awt[billymg]: Currently working on SelfChain fork notifications, but fiat work is bogging me down for a couple days.
asciilifeform[billymg]: awt: when you get a chance, 1-at-entry-per-peer plox
asciilifeform[billymg]: (or is that already in next ver.?)
awt[billymg]: asciilifeform: in 9984 - you may need to repeer anyone not actively broadcasting rubbish to remove their old AT entries.
asciilifeform[billymg]: awt: a ok. (maybe can manually clean via sqlite console..?)
awt[billymg]: asciilifeform: this also
shinohai[billymg]: good day pestizens
awt[billymg]: lol good day shinohai
awt[billymg]: I guess pestizens sounds better than pests
shinohai[billymg]: lmao
shinohai[billymg]: "Hey folks, I got blatta working but it keeps crashing. Do you think it's because I'm trying to pest on a RAID array?"
shinohai[billymg]: ba dum TISS
awt[billymg]: lulz
PeterL[billymg]: pesters? pestians?
awt[billymg]: Shit. Had a dupe come through via asciilifeform with 2 bounces. Original hash was 7990fd9358211, dupe hash is bd5d37cec67f.
awt[billymg]: Possibly because someone is not upgraded. Might need finer version granularity than pest versions in the message version field due to pest implementation versions being a separate thing.
PeterL[billymg]: maybe a separate message type where you can ask a peer what version they are using? Or you could just ask people what version they have instead of baking it into the program?
jonsykkel[billymg]: if the speaker padded with ' ' instead of \0 thing was fixed that would give "same" message diffrent hashes
PeterL[billymg]: jonsykkel: but the message should not be getting repacked by peers, it should just be passed on with whatever padding the original creator put in it
jonsykkel[billymg]: PeterL: indeed, dunno what code does
PeterL[billymg]: I think the packing/repacking problem got removed in 9988?
jonsykkel[billymg]: could be
jonsykkel[billymg]: maybe some peer is still repacking
asciilifeform[billymg]: jonsykkel, awt : hash is over ~entire~ field, not simply the portion prior to terminator
asciilifeform[billymg] thought it was 100% clear from spec?
jonsykkel[billymg]: yes, thats what i mean, spaces as terminator insted of \0 would give diffrent hash and fall through working deduper
jonsykkel[billymg]: just guessing at whats happening anyway
asciilifeform[billymg]: prolly oughta prohibit trailing spaces, in Text, fwiw
PeterL[billymg]: why would trailing spaces be bad?
PeterL[billymg]: hmm, maybe that is why, just got a dupe of my message (had a couple spaces after ?)
PeterL[billymg]: should that get stripped by the pestron before packing the message?
shinohai[billymg]: YEah I see PeterL 's last message twice, dupe wat as [awt]
PeterL[billymg]: I feel like I should be able to end a line with a space if I want to
asciilifeform[billymg]: in either case you dun want to modify message in any way when rebroadcasting
jonsykkel[billymg]: gotted this now http://zzz.st/up/Z8JImRDO/ so if that 8c1cb1... is the message hash what i guessed isnt whats hapening
asciilifeform[billymg]: (loox like currently gets mutilated)
asciilifeform[billymg]: when you rebroadcast, same 428 bytes which came in, oughta come out, per spec
asciilifeform[billymg]: (best not to even try to 'repack' or whatver similar process, that oughta happen only when originating a msg)
awt[billymg]: I saw another dupe via asciilifeform which I then rebroadcast
awt[billymg]: This is the dupe: 07aa5cf7b1
awt[billymg]: had 1 bounce
asciilifeform[billymg]: awt: does the current version mutilate payloads as described above? (or, is my station simply relating mutilations from somebody using earlier rev.?)
asciilifeform[billymg]: *relaying
jonsykkel[billymg]: i got dupe in irc that was not mutilated according to message hash in log
jonsykkel[billymg]: so it seems deduper is at fault?
awt[billymg]: asciilifeform: should pass the message bytes unmodified. apparently not doing that.
awt[billymg]: jonsykkel: messages are not added to the dedup queue until after the embargo period in blatta. So you could see two identical messages logged.
jonsykkel[billymg]: awt: i saw them in irssi though
awt[billymg]: jonsy
awt[billymg]: jonsykkel: are you certain there are no other instances of that message in your logs?
jonsykkel[billymg]: ill check
jonsykkel[billymg]: ah sure is
jonsykkel[billymg]: my bad, i guess messages are getting mutilated after all
awt[billymg]: jonsykkel: yeah it appears that my station mutilated the message in this case: http://paste.deedbot.org/?id=Se6z
awt[billymg]: A Mutilator Mystery, by awt
jonsykkel[billymg]: should do hexdump and see wahts going on
jonsykkel[billymg]: christmas mystery
awt[billymg]: Version in progress has a refactoring that makes the path of forwarded messages somewhat more clear.
PeterL[billymg]: is everybody on this test net down to 9985 or is somebody still running an older version?
shinohai[billymg] is on 9985
jonsykkel[billymg]: 9985 bros
awt[billymg]: yeah looks like the bug wasn't eliminatedin any version