Logs for #eulora

Go to: #trilema #pizarro #asciilifeform #trilema-mod6 #chainstate #eulora

2018-7-30 | 2018-8-1

mircea_popescu: Mocky not afaik. your damned client.
mircea_popescu: advise-wise i'd just keep the current dependencies for now, because on something this large it's easier to replace by pieces. but that's as far as that goes.
Mocky: ok, just thinking things through, seeing if I've grasped what's involved. seems my employer is circling the drain and I have some decisions to make
mircea_popescu: i c.
mircea_popescu: prolly much safer to go by parts.
Mocky: yeah and with the comms part probably the first thing, right?
mircea_popescu: probably, yeah.
mircea_popescu: truth be told there's A LOT one could do about extant client to make it better, many of which quite low hanging.
mircea_popescu: and it's, intelligently, a very convenient moment to move. land rush so to speak ; as republic is evidently not going to recognize "intellectual property" of anyone outside the wot itself, can very practically homestead. bring improvements and there you go, own the indian's land, be part of a new civilisation.
Mocky: that's a good point
mircea_popescu: history repeats while rhyming.
lobbesbot: A#338 O=1bn HB=1bn E=08-03 16:33:06 (86h18) >>> 8.2k wFF q0
lobbesbot: A#342 O=15.5mn HB=15.5mn E=07-31 10:52:16 (8h37) >>> 240450 SR q68
lobbesbot: --- end of auction list, 1.0155bn total bids ---
lobbesbot: AUCTION # 342 has ENDED: 240450 SR q68 SOLD to mircea_popescu for 15.5mn coppers. Attn: diana_coman
lobbesbot: A#338 O=1bn HB=1bn E=08-03 16:33:06 (77h18) >>> 8.2k wFF q0
lobbesbot: --- end of auction list, 1bn total bids ---
Mocky: got disconnected in the night while bot mining. apparently now table is gone
diana_coman: o.O Mocky did the bot go away from the table/fell down or anything?
diana_coman: disconnect in itself does not make the table vanish
Mocky: na, still standing the same spot when i got back in. was in the middle of stacking output to table when disconnect.
diana_coman: uhm, kind of... impossible, lol
diana_coman: unless someone else took it while your char was dc perhaps
diana_coman: if it's guarded by your char, it stays guarded (i.e. won't vanish) even if you disconnect
Mocky: yeah was quite surprised not to see it there when i came back in
diana_coman: fwiw bot behaved quite well for me so far, it even changed tool fine; it just runs out of space/lbn mostly but that's not it's fault
Mocky: i'm glad to hear it
Mocky: feels kinda weird tbh, been running this thing every day and night for weeks constantly finding new problems or regressions. now just seems to work, lol
lobbesbot: A#338 O=1bn HB=1bn E=08-03 16:33:06 (74h18) >>> 8.2k wFF q0
lobbesbot: --- end of auction list, 1bn total bids ---
Mocky: so i can get flotsam at q45 now if i put 2 cft in my small bundles but I'd have to sell at > 132% to make any profit. which isn't bad considering i paid 135% for cft and 156% for tools
Mocky: but also i only get them 15 at a time so like ~60 bot hours per 100k
diana_coman: oh hey, q45 on flotsam is not bad
diana_coman: Mocky, so you feel better: bot got stuck with "cannot pickup table, stopping" ; however, main window says "that item cannot be picked up" so apparently it was trying to pickup something other than table?
diana_coman: fwiw I picked it up manually no problem too
Mocky: so much better :)
diana_coman: heh; let me know if there's any other info useful to track this down; it was after mixing, it said "Moving" and then "Picking up table" "54 attempts done, 4946 attempts left." "Cannot pickup table, stopping"
Mocky: for that i would normally consult network log to see if the eid bot was using for table was correct and if so were there other entities nearby with same eid
Mocky: but since bot mixed output to table, doesn't sound like normal table glitch i've seen
Mocky: prolly not anything else useful you could tell me
diana_coman: does bot use specific move item cmd or just generic pickup?
diana_coman: iirc pickup uses target
diana_coman: so if target gets messed up, it will fail?
Mocky: no, it directly reads persist_item msgs from server and stores eid of table. then uses `/pickup eid:1234`
diana_coman: hm, weird
Mocky: just searched thru my old logs, 10 most recent occurrences of 'that item cannot be pickup up' are all dupe eids like: http://p.bvulpes.com/pastes/uMmud/?raw=true
Mocky: notice 1st and 4th lines
diana_coman: hm, but it moved stuff correctly to it; weird
Mocky: unfortunately that '/pickup'Error: Not Decoded
Mocky: always logs when using `/pickup eid:1234`, if success or no
Mocky: maybe moving brought it closer to other entity
Mocky: when my table got same eid as my player the other day, was able to pickup manually after moving around a bit
diana_coman: ah, could be it got a new entity with same id as table, hm
diana_coman: ftr I did not move anywhere to pick it up
lobbesbot: A#338 O=1bn HB=1bn E=08-03 16:33:06 (73h18) >>> 8.2k wFF q0
lobbesbot: --- end of auction list, 1bn total bids ---
Mocky: yeah that bit is definitely weird
Mocky: for completness, bot just stopped for me cuz 'looks liek table has already been dropped, but i'm still overweight, oh noes!'
diana_coman: ahahha
Mocky: btw, table wasn't actually dropped
diana_coman: so then how it figured out it was?
diana_coman: does it check inv for that id?
Mocky: it searches inv for tables, keeping in mind how many tables were present at the start. if there is exactly one less table than at start, it figures one was dropped
Mocky: unfortunately this is on a test system, not on my dev system and i don't have logging enabled so can't research why. will have to wait for it to recur on dev
diana_coman: for another potential bug with this approach: does it take into cons stacked tables?
Mocky: oh no, looking at my bot output window, it was trying to do things out of order. 'already dropped' messages happening before the end-of-round status message, and then 're-dropping glitched table' without having tested the table
Mocky: it doesn't consider stacked tables at the start or during. so if you unstacked them manually during a run, i would expect bot to tilt. otherwise i think it would be ok
Mocky: i need a way to manually induce lag when testing
Mocky: i think a spike of lag waiting for table pickup to confirm is my problem, and i put in one fix for it last week, but I can't actually test it on command
Mocky: anybody got a eulora bouncer sitting around, lol
diana_coman: I started it once with some stacked tables and it messed them around (though I didn't quite bother to fully follow what it did - it seemed at least there still were as many as I started with, just not stacked up anymore)
Mocky: never even thought to test that!
Mocky: i better check that out while i still have 2 tables, lol. i lost one last week too when bot walked off a cliff
diana_coman steps into "weird test cases" naturally 
diana_coman: lol! I'll give you some free tables next time I'm in town
lobbesbot: A#338 O=1bn HB=1bn E=08-03 16:33:06 (72h18) >>> 8.2k wFF q0
lobbesbot: --- end of auction list, 1bn total bids ---
Mocky: so which things do you see as hanging low? http://logs.minigame.biz/2018-07-31.log.html#t01:41:24
lobbesbot: Logged on 2018-07-31 01:41:24: <mircea_popescu> truth be told there's A LOT one could do about extant client to make it better, many of which quite low hanging.
mircea_popescu: Mocky take the complaints in logs as a guide say. "i logged in and my client thinks i'm x"./
mircea_popescu: take all the "ghosting"
Mocky: ok, yeah
mircea_popescu: and odds are, once a thread or two are pulled...
mircea_popescu: tell me, why the fuck does this client even need 2gb ram ?
mircea_popescu: i mean, i had to buy more ram because of this wtf.
lobbesbot: A#338 O=1bn HB=1bn E=08-03 16:33:06 (71h18) >>> 8.2k wFF q0
lobbesbot: --- end of auction list, 1bn total bids ---

2018-7-30 | 2018-8-1