Show Idle (>14 d.) Chans

← 2021-11-30 | 2021-12-02 →
00:01 awt asciilifeform: yep
00:01 asciilifeform possibly this was from having 2 stations on 1 box tho
00:01 asciilifeform ( pestbot ? )
00:02 awt Ha now this is interesting - pestbot station didn't get the log quotes
00:02 asciilifeform oh nm pestbot is awt's thing innit
00:02 asciilifeform log was from billymg
00:02 asciilifeform oh and here's my 1st console dupe
00:02 asciilifeform 'asciilifeform[billymg] | log was from billymg '
00:03 awt asciilifeform: I saw the dupe at one station, but not the other
00:03 asciilifeform deduper clearly b0rken
00:03 asciilifeform tho must be in a subtle way where only occasionally barfs
00:03 awt yep
00:04 awt
00:04 bitbot Logged on 2021-11-30 20:10:48 awt[billymg]: asciilifeform: when you see that's likely the irc client sending or receiving something to or from the local pest station
00:09 awt odd. my main station is not passing on the log quote message even though bounce count is 1.
00:09 awt
00:09 bitbot Logged on 2021-11-30 20:10:48 awt[billymg]: asciilifeform: when you see that's likely the irc client sending or receiving something to or from the local pest station
00:11 awt
00:11 bitbot Logged on 2021-11-30 20:35:50 billymg: neat
00:12 awt pardon the exessive noise while I debug this
00:13 awt *excessive
00:16 awt
00:16 bitbot Logged on 2021-11-30 20:35:50 billymg: neat
00:22 awt
00:22 bitbot Logged on 2021-11-30 20:35:50 billymg: neat
00:22 awt ok figured it out. was not rebroadcasting simple hearsy.
00:22 awt *hearsay
00:34 billymg i noticed this happening earlier too
00:34 bitbot Logged on 2021-11-30 23:17:47 awt[billymg]: ah wait it did -- just took a long time to show up on the web
00:34 bitbot Logged on 2021-11-30 03:07:36 billymg: ah, nm, just a delay
00:46 awt
00:46 bitbot Logged on 2021-11-30 20:35:50 billymg: neat
00:47 billymg regarding pestbot, wouldn't i at least need to be peered with it in order for its messages to make it to bitbot?
00:48 awt billymg: no. only 3 hops
00:48 billymg ah ok
00:54 awt
00:54 bitbot Logged on 2021-11-30 20:35:50 billymg: neat
~ 17 minutes ~
01:12 billymg try saying something with pestbot again
01:13 pestbot pesting testbot
01:13 billymg ok, made it into the logs that time
01:14 billymg after i did `/peer pestbot`
01:14 billymg (from this station, not the bot)
01:17 billymg that behavior is consistent with what i saw before as well, e.g. when jonsykkel joined i had to peer with him from this station first before bitbot would see any of his messages
01:18 awt billymg without a key for pestbot you won't be able to decrypt immediate messages from pestbot
01:18 awt the problem was with blatta - it wasn't rebroadcasting immediate messages - only hearsay messages
01:19 awt sorry let me rephrase that
01:19 awt it wasn't rebroadcasting simple hearsay messages
01:19 awt so any message your main station receives from pestbot is simple hearsay
01:20 awt because you have no key for pestbot
~ 26 minutes ~
01:46 billymg awt: lost connection on my main pest station again, with this sqlite error
01:47 * billymg suspects his last few messages after never made it to anyone
01:47 bitbot Logged on 2021-12-01 01:20:01 awt[billymg]: because you have no key for pestbot
01:52 billymg << i understand that the messages will still be hearsay until you have also keyed the peer, but for whatever reason, a simple `/peer` (without adding a key) is what it takes to pass the messages along to bitbot
01:52 bitbot Logged on 2021-12-01 01:19:45 awt[billymg]: so any message your main station receives from pestbot is simple hearsay
01:53 asciilifeform btw this aint per spec : simply /peer should do absolutely nuffin without a keying
01:53 asciilifeform (prolly oughta put this in the notes)
01:53 asciilifeform a peer without a key is not a peer.
01:54 asciilifeform ditto if there is no AT entry
01:55 asciilifeform possibly /peer oughta require a key as a 2nd argument, the current way may be confusing
01:56 billymg i think that makes more sense
01:56 asciilifeform (if there is no AT entry, one might still ~receive~ a msg from the peer tho, even if cannot yet send)
01:57 asciilifeform w/out any keys, cannot receive or send.
~ 24 minutes ~
02:21 awt billymg: your adding peer is not the only thing that's different. I'm now running a station with a bunch of fixes
02:22 awt incidentally the sqlite3 error you pasted is due to blatta implementing threading badly
02:23 awt I think I'm going to have to have a single thread that does all db access, and just performs operations from a queue. right now occasionally two threads attempt to update the db at the same time
02:30 pestbot
02:30 bitbot Logged on 2021-11-30 20:10:48 awt[billymg]: asciilifeform: when you see that's likely the irc client sending or receiving something to or from the local pest station
02:31 awt heh billymg gets a notification everytime someone quotes a line
02:32 awt or wait - no, not in his console, since it's an immediate message. nm.
~ 1 hours 17 minutes ~
03:50 billymg awt: heh, yeah, i do get the notifications on all bitbot echos since i'm the intermediary on them
03:55 billymg awt: these are the loglines in my weechat console, the "pestbot (pestbot) has joined #pest" occurred immediately after running `/peer pestbot`
03:55 awt yeah blatta sends a join message when you peer someone so they show up in your nicklist
03:55 awt also btw we can test your peer theory if you unpeer
03:56 awt if you unpeer pestbot, I mean
03:56 billymg right, i'll try that
03:57 billymg i'm just looking at the timestamps and at 18:53 the pestbot[awt] message didn't get logged (same with the previous messsages from earlier), then at 19:09, after i peered, it worked
03:57 billymg lemme try unpeering
03:57 billymg ok, got the "pestbot has left" message
03:58 billymg yeah, not in the logs
03:58 awt indeed
03:58 billymg ok, will peer again
03:59 billymg peered
03:59 pestbot testing testing
03:59 billymg yep
03:59 awt that's so not cool
04:00 awt because pestbot gets bitbot messages w/o peering
04:02 billymg so pestbot could see bitbot messages (i'm guessing via billymg?) even when i was not yet peered with pestbot?
04:02 awt not sure
04:02 asciilifeform btw before i fughet, awt : this aint in spec yet, but pestron oughta rejects msgs with higher version than itself
04:02 awt asciilifeform: yes indeed
04:03 awt I should add that rather soon to blatta
04:13 awt ok well for the last dupe I saw, I can confirm via the logs that the message hash reported on the originating station is different than on the station that sent back a dupe.
04:14 awt WTF? message hash getting fucked right in front of my eyes
04:24 awt ah ok I think I might know why peering/unpeering even w/o a key is affecting what bitbot sees. There is some logic still in your version that checks if a messages is in-wot. according to this logic the message would be in-wot, even w/o a key.
04:25 awt and only in-wot hearsay is getting rebroadcast in your version
04:25 awt jesus
~ 34 minutes ~
04:59 pestbot testing potential dupe fix
04:59 awt Hopefully shouldn't see any more dupes oritinating from my main station or pestbot.
05:12 awt eliminated recompiling received broadcast messages - just passing the bytes straight on through now.
~ 8 hours 34 minutes ~
13:47 awt Now testing a version that is hopefully somewhat smarter about locking around writes to the db.
13:49 pestbot testing from pestbot
~ 42 minutes ~
14:31 asciilifeform aaand running awt's new thing
14:33 asciilifeform anybody tuned in ?
~ 16 minutes ~
14:49 awt asciilifeform: here
14:49 asciilifeform a++
14:50 pestbot Pestbot also here
14:52 awt Shit. Still getting threading related exceptions.
14:53 PeterL I just upgraded (or is it downgraded since the numbers are decreasing) blatta to 9986, lets see if it works ...
14:53 jonsykkel 123
14:53 awt Hello PeterL and jonsykkel
14:53 jonsykkel hi
14:53 jonsykkel oh bot sees me now
14:54 jonsykkel seems like im only receiving packets from peterl
14:54 awt Yeah I'm not getting anything from you
14:55 jonsykkel weird
14:55 PeterL every time I connect, in my IRC client console I get a "ERROR: USERHOST: Unknown command", any ideea what is causing that?
14:55 awt jonsykkel: this you?
14:56 jonsykkel awt[PeterL]: thats me
14:57 awt PeterL: that's because Blatta doesn't support that command, I think
14:57 awt jonsykkel: showing up as a martian packet
14:57 jonsykkel awt: key starts with YOZ?
14:57 awt jonsykkel: yes
14:58 jonsykkel i have same then likely
14:59 awt Looks like possible issue with automatically updating the at when there is no address set for a peer.
15:03 awt The minute I note that there haven't been any dupes, there will be dupes.
15:04 awt or not
15:16 asciilifeform
15:17 awt asciilifeform: yep likely threading issue.
~ 1 hours 27 minutes ~
16:44 billymg when i see "PeterL[asciilifeform|awt]" does that mean hearsay via asciilifeform who got it hearsay via awt (or perhaps the other way around)? or does it mean hearsay from both asciilifeform and awt, independently?
16:45 billymg btw awt, typo in your latest post: first sentence refers to the previous version
16:47 PeterL billymg: that means you are seeing things from me passed to you by both asciilifeform and awt independently
16:48 billymg PeterL: got it, thanks
16:50 awt billymg: ty fixed
16:50 billymg i'm also wondering what to do in the logger with the hearsay annotations in the future. should log bots just peer with everyone in the net? should it just be stripped off, or maybe shown on hover? there's also the issue that [] are valid characters in IRC nicks
16:51 billymg as in, potential issue when it comes to stripping the annotations
16:51 awt billymg: this is a good question
16:52 awt if we leave the annotations in, everyone's log is going to look slightly different
16:53 awt people might end up follwing the logger that has the most peerings for those speaking in the log
16:53 billymg yeah, that's kind of how i see it, the logger will have to be very liberal with its peerings
16:55 billymg i've already been thinking of potential frontend knobs for the logger, ways for the person browsing the logs to customize their view to some extent
16:56 billymg one idea i had was for a simple ?highlight=blah param that could be used to track words (potentially your own nick, to simulate mentions) in a thread
16:57 billymg thinking of the pest model i could also a see a way to "mute" nicks as being useful
16:58 billymg so then a log bot might cast a pretty wide net, and log readers could trim it down to the speakers they're interested in
16:59 awt billymg: are you on 9986 btw?
16:59 billymg not yet
17:00 awt ah ok. well, it would be good to run to see if it eliminates the dupe issue, but otoh it crashes frequently due to changes I made wrt threading.
17:00 billymg the dupe issue hasn't been so bad on my end, only noticed it two or three times
17:01 billymg saw the comment in #a about the threading issue, is that similar to my sqlite error i posted?
17:01 awt billymg: yes
17:03 billymg ok might just wait for the next release then, but will update if you need me to test anything on 9986
17:04 awt sounds good. if there are dupes and they only come through your station that would also be good info
17:06 billymg got it, will report
~ 1 hours 19 minutes ~
18:25 asciilifeform hmm my pestron not wedged again yet since restart. i suspect will be tricky to reproduce 'heisenbug'
~ 54 minutes ~
19:20 awt Trying out a patch that uses new cursors for each method in
19:21 awt Conflicts seem to happen way more often here than on my local testnet.
~ 1 hours 48 minutes ~
21:09 awt asciilifeform: so far so good with the cursor patch I'm trying. Will let it run another 20 hrs or so.
← 2021-11-30 | 2021-12-02 →