Day changed to 2024-12-01
[02:02] jfw: it looks like it's indeed a much more recent shift on the enterprise HDD side, with datasheets for older models indicating 512-byte sectors (with 520 or 528 byte formatting options) without mention of some larger underlying; this includes the 4TB 'Constellation ES.3' set we got before; more current seagate products in lower capacities show '512N SAS' as a format option, suggesting 'native' as
[02:02] jfw: opposed to the new 512e term. but the peak there seems to be Exos 7E10 6TB, which isn't showing up for sale on their own store or the usual platforms.
[02:28] jfw: yet another speedbump is the cabling: the breakout 'squid' cable from the 3ware controller terminates in SATA type connectors (with the L-shaped edge keying), which accomodates the supermicro backplane which passes it through to the full SAS drive connection including power
[02:30] jfw: but for individual SAS drives in a workstation without the backplane, it seems an adapter is needed because neither SATA data nor power connectors will fit directly onto the SAS drive.
[02:33] jfw: and for added excitement, with this 'cable management' style that all the ATX cases switched to, there's little room for connectors coming out of the drive as they're positioned right against the back panel of the case. would need to check if I can mount them in the opposite direction.
[02:39] jfw: https://www.amazon.com/EDIMS-Internal-SFF-8087-SFF-8482-29Pin/dp/B07HL4M1P2/ looks like the ticket
[03:06] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12736 - for completeness, I did find the nominally 512N variant of that 6T Exos drive with SATA interface (model ST6000NM000B), at CDW.
[03:06] sourcerer: 2024-12-01 02:02:49 (#jwrd) jfw: opposed to the new 512e term. but the peak there seems to be Exos 7E10 6TB, which isn't showing up for sale on their own store or the usual platforms.
[03:10] jfw: so it seems I face an entirely underinformed decision of whether I care most about interface protocol elitism, sector size traditionalism, or raw capacity.
[03:13] jfw: my bias is toward the original plan of ~10T SAS. the sector size thing looks like a lost cause with just this one specific item available today and probably not even in production anymore
[03:20] jfw: western digital's site won't even load a product catalog, with scripts
[03:46] jfw: https://www.sqlite.org/psow.html - some evidence of the sadness of big sectors.
[03:47] jfw: along with https://www.sqlite.org/atomiccommit.html
[03:55] jfw: I don't know of a comparable explanation for MySQL, but there's indication that some operations are not 4k aligned
[06:37] jfw: dorion: new plan: open up the machine tomorrow to check on the question of space for SAS connector (as is or by drive reversibility); if all is well, order 2 more of *the same 4T seagate drive as before* (found it for $90). going all the way on the 'standardize' direction. basically, we know we have database applications, some of which may be write-intensive. we know we need reliability (and have
[06:37] jfw: had various troubles with SATA stuff in the past). whereas we don't as yet have any projects that require that many terabytes; if we take some on, they can better inform the specific hardware needs if indeed there's no longer such thing as the best all-around drive. maybe we'd want something even bigger, or more options come on the market for sector size (one can dream)
[19:48] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12739 - just from checking the manual illustrations I'm recalling the problem at least with this case: it uses slide-out plastic caddies to mount the 3.5" drives, and even if it's physically possible to attach the drive with ports facing the front, the cover plate of the caddy will block the cables.
[19:48] sourcerer: 2024-12-01 02:33:57 (#jwrd) jfw: and for added excitement, with this 'cable management' style that all the ATX cases switched to, there's little room for connectors coming out of the drive as they're positioned right against the back panel of the case. would need to check if I can mount them in the opposite direction.
[19:52] jfw: the only things I'm turning up with right-angled connectors on the SAS drive end (SFF-8482) are right-angle adapters that turn it into SATA type connectors, which is annoying but might work - use the current breakout cable with its SATA connectors, then molex to SATA power adapters as necessary.
[19:55] jfw: the prior linked breakout cable would have done exactly the desired thing without further adapting... but what can you do.
[20:03] jfw: but the right angle could create its own problems with fitting the drives next to each other. so I guess it's still worth checking if those caddies can be sawed open or something.
[20:16] jfw: altogether, computer assembly is starting to look a lot like contemporary software development: plenty of feature-filled frameworks around which economy compels you to adopt and then you spend all your time tracking down their bugs or working around their spurious limitations.
Day changed to 2024-12-02
[00:49] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12755 - so, the only way a drive will go reversed into the caddy is by fully removing at least the clips that snap the caddy into the chassis. there aren't enough screw holes to mount the drive 'manually' without the caddy, but it looks like there's at least one usable to secure it from sliding if the clips are removed.
[00:49] sourcerer: 2024-12-01 20:03:36 (#jwrd) jfw: but the right angle could create its own problems with fitting the drives next to each other. so I guess it's still worth checking if those caddies can be sawed open or something.
[00:52] jfw: there's enough clearance between drives that the right-angle MIGHT work, and enough to the back case wall that the straight connectors MIGHT fit with some gentle cable pre-bending. so it seems clear enough to go ahead with SAS, but will have to grab 'one of each' on the cabling and see what works.
[04:52] jfw: order placed - most of it should be in miami by next tuesday the 10th. by which point I expect the forwarders will be in holiday backlog and maybe it arrives by new year's.
[04:55] jfw: I went with a slightly updated drive model, ST4000NM0025 which supports the new 12 Gbit SAS link speed, advertises slightly better reliability numbers, higher sustainable throughput and lower power draw, and maintains 512N sector format.
[13:54] dorion: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12756 -- pretty sad, but thanks for wading through the swamps.
[13:54] sourcerer: 2024-12-01 20:16:49 (#jwrd) jfw: altogether, computer assembly is starting to look a lot like contemporary software development: plenty of feature-filled frameworks around which economy compels you to adopt and then you spend all your time tracking down their bugs or working around their spurious limitations.
[13:54] dorion: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12759-- sounds good, I'll keep an eye out for the package.
[13:54] sourcerer: 2024-12-02 00:52:21 (#jwrd) jfw: there's enough clearance between drives that the right-angle MIGHT work, and enough to the back case wall that the straight connectors MIGHT fit with some gentle cable pre-bending. so it seems clear enough to go ahead with SAS, but will have to grab 'one of each' on the cabling and see what works.
Day changed to 2024-12-04
[16:57] nekoluce: hello
[17:46] dorion: hello nekoluce, welcome. who might you be ?
[18:22] jfw: Luz found her way in.
[18:26] jfw: nekoluce: I remembered the Android IRC client I had working once, it's Atomic.
[18:26] jfw: now we just need that bouncer setup...
[18:29] jfw: dorion: I was thinking of posting a bounty for that, not because it's hard but because it's relatively self-contained, easy to specify and verify, in the spirit of offering ways in
[18:31] jfw: basically, take your pick of Gales or CentOS, pick the software, version & dependencies, write the recipe to build from source, install, configure, and create user IDs
[18:33] jfw: maybe .002 BTC or you think that's too rich?
[18:36] jfw: last night's quit/join noise btw was from a planned router maintenance at hetzner.
[18:39] jfw: I need to fix our bots to rejoin properly - wotbot has the code working fine to detect timeouts but it just needs to be run in a service to auto-restart; sourcerer is the awk based prototype and doesn't detect the server's absence; the thought was to implement the logging stuff in the new 'production' framework
[18:40] jfw: wouldn't mind paying to get that done either but it's probably a higher code-learning investment up front
[19:00] jfw: hi nekoluce, here's an example of the use of the log and quoting: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12769
[19:00] sourcerer: 2024-12-04 18:26:40 (#jwrd) jfw: nekoluce: I remembered the Android IRC client I had working once, it's Atomic.
[19:03] nekoluce: "Quoting" ?
[19:03] jfw: yes, see how I just pasted the link and the bot replied with the referenced line
[19:04] nekoluce: i'm trying to click on the link but it doesnt seem to work .
[19:05] jfw: maybe right clicking?
[19:07] nekoluce: it appears as the same website ?
[19:08] nekoluce: is it normal ?
[19:08] jfw: it should highlight and scroll to the specific line
[19:08] nekoluce: it is highlighted on the website but not on lime
[19:09] nekoluce: did you arrive safe ?
[19:10] jfw: barely; I was attacked by a row of hungry ducklings.
[19:11] nekoluce: you know i do have a park of hungry ducklings nearby my home jaja.
[19:11] nekoluce: there is even a lake
[19:12] jfw: nice, we usually get some families near my parents' place in Maine
[19:13] nekoluce: What is Maine like ? Cold or warm ?
[19:14] jfw: seasonal; it's pretty far north, on a level with southern Canada, so long winters but temperatures moderated by the sea. summers are historically quite pleasant but last few years have been hot.
[19:15] nekoluce: hot as panama ? it seems beautiful and lovely scenery .
[19:15] jfw: it can be, but not as humid.
[19:15] nekoluce: wish someday I meet Maine owo
[19:16] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12786 - that makes sense, it's just a feature of the web view. but that's why the bot quoting helps to follow right from the client.
[19:16] sourcerer: 2024-12-04 19:08:50 (#jwrd) nekoluce: it is highlighted on the website but not on lime
[19:16] nekoluce: hi sourcerer !
[19:25] nekoluce: it has a password for atomic
[19:26] nekoluce: sorry for not saying hi rob
[19:49] jfw: like a password on startup, really? odd since it's supposed to be open source. did you grab it from play store or f-droid?
[19:53] jfw: I have version 2.1
[20:08] nekoluce: from fdroid
[20:17] nekoluce: oh it work
[20:17] nekoluce: but i have another username
[20:17] nekoluce: so I have now two ?
Day changed to 2024-12-05
[03:48] jfw: the fiat world hits a new benchmark low with one USD now fetching less than 1000 satoshi.
[03:51] lru: and I finally bought some
[03:52] jfw: I guess it's time for the correction then!
[03:52] lru: lol
[03:52] lru: sorry everyone :-)
[03:52] lru: j/k
[03:53] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12807 - an irc nickname can only be held by one connection at a time. so you can set the second client to use a variant of the name, or just use one at a time. I think this would also be improved with a bouncer setup
[03:53] sourcerer: 2024-12-04 20:17:44 (#jwrd) nekoluce: so I have now two ?
[03:53] jfw: well congrats on dipping the next toe in, lru.
[03:54] lru: thanks
[03:55] lru: how do your bounties work btw? is it first person who completes gets paid, or first person who accepts the challenge and then completes in reasonable time?
[03:57] jfw: tbd, but in this case I'd be inclined to the latter
[03:58] jfw: i.e. it's a commitment and what you make of it will reflect one way or another
[03:59] lru: sounds good... I suppose if there is a crowd wanting to participate, nothing stopping people splitting the reward and working together
[04:00] jfw: WoT identity is a must, needless to say. and sure, it might be delegated but someone in particular will be responsible
[04:06] lru: I see
[04:06] lru searches for wotbot docs
[04:10] lru: found 'em
[11:17] Atomic_0TTFW: good morning 🌅
[11:18] Atomic_0TTFW: ohayogozaimasu ! desu
[14:46] nekoluce: hello mr xissburg
[15:09] xissburg: nekoluce: neko ga imasu!
[15:10] nekoluce: how is your day ?
[15:10] xissburg: glorious day
[15:11] nekoluce: amazing having a nice weather.w. ?
[15:14] nekoluce: i started using irc yesterday is quite fun
[20:20] jfw: what's this, secret codes? unfair!!
[20:25] jfw: I'm getting "good morning! it is" and "there is a cat"
[20:42] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12773 - I picked that number because it's about what I'd want if doing it for someone else; someone less experienced or familiar with the environment would take longer and get less per hour, so that seems fine; and if it's "too much" then it could still be a good idea for now, viewing the excess as publicity expense. if at some point we end up with
[20:42] sourcerer: 2024-12-04 18:33:04 (#jwrd) jfw: maybe .002 BTC or you think that's too rich?
[20:42] jfw: more qualified applicants than we know what to do with, then we can look at competitive bidding
[20:46] jfw: xissburg, you got any war stories from hiring engineers or gig workers generally?
[20:50] nekoluce: hey are you guys going to the btc party ?
[20:50] xissburg: jfw: hm not really, just firing due to incompetence
[20:50] xissburg: do you?
[20:51] xissburg: nekoluce: nah
[20:52] jfw: xissburg: just dipping my toes in now
[20:52] nekoluce: having a bath ?
[20:52] jfw: nekoluce: heh, not literally
[20:53] nekoluce: ah xD
[20:53] jfw: what's the btc party you mention?
[20:54] jfw: the offchain guys doing something?
[20:54] nekoluce: tonight there is a btc party at
[20:54] nekoluce: yes the offchain guys
[20:54] nekoluce: like three groups together
[20:55] nekoluce: the biergarten
[20:55] nekoluce: since 6
[20:55] nekoluce: pm
[20:55] jfw: perhaps I'll stop by
[20:55] nekoluce: yay. !
[20:56] xissburg: jfw: I think it's good to be ready to fail fast/fire fast
[20:58] jfw: right, basically you make yourself vulnerable to some extent so see what people do with that and don't extend more credit than necessary
[20:59] nekoluce: thank you xissburg ,bye bye Ill go get ready
[21:02] jfw: xissburg: any particular venues you've found good for announcing offers, or do you find the word spreads itself if the offer is good?
[21:17] xissburg: jfw: you have to announce it where the likely candidates hang out
Day changed to 2024-12-06
[16:40] nekoluce: Good morning !
[18:27] jfw: good day nekoluce
[18:28] nekoluce: is it rainy at your house ?
[18:30] jfw: yes, a light rain
[18:32] nekoluce: rain is lovely makes panama less hot and more cloudy those days are fun with hot cocoa .
[18:33] nekoluce: your welcome
[18:34] jfw: I find hot cocoa improves just about any day indeed.
[18:34] jfw: and I like the rains here except for those times when I'm on foot and the roads are flooding.
[18:35] nekoluce: oh ,yes .
[18:35] jfw: just a different kind of adventure though I guess.
[18:36] nekoluce: btw there is a guy in one of the groups promoting an event maybe it could help get exposure for you guys toward more people in the market is called Panama Future .I dont know if you are interested ?
[18:38] jfw: couldn't hurt to have a look, does it have a site/article/anything?
[18:38] nekoluce: actuaally there are two
[18:38] nekoluce: Panama Blockchain week also is another one
[18:38] nekoluce: let me look for it
[18:41] jfw: xissburg: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12862 - makes sense, so figuring that out is part of the challenge
[18:41] sourcerer: 2024-12-05 21:17:57 (#jwrd) xissburg: jfw: you have to announce it where the likely candidates hang out
Day changed to 2024-12-07
[16:37] nekoluce: hello
[21:31] jfw: hi nekoluce
[21:32] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12873 - I guess that would be https://panamafuture.com/
[21:32] sourcerer: 2024-12-06 18:36:53 (#jwrd) nekoluce: btw there is a guy in one of the groups promoting an event maybe it could help get exposure for you guys toward more people in the market is called Panama Future .I dont know if you are interested ?
[21:34] jfw: I'd see that as scammy until proven innocent. blockchain tokens for real estate? property rights in real estate are ultimately arbitered by the ruling powers on the ground, while the block chain exists outside of any particular state. those mix like oil and water.
[21:35] jfw: serve up some marketing word salad and it'll provide the dressing, sure.
[21:47] jfw: this might mark a key difference between us and perhaps noisier crypto bros. at that party the other night they went around sharing, almost boasting it seemed, about their biggest losses, how they've been robbed but keep getting back up. taking risks is necessary to get anywhere, sure - but get educated and take smart risks.
[21:47] jfw: "don't get robbed" is pretty central to my recipe for wealth, as I talked about for instance here.
[21:52] jfw: someone taking jwrd's advice at that time is up 2.5x today or about 38% yearly without leverage.
[21:55] jfw: in dollar terms, sure, for the sake of comparing investments.
Day changed to 2024-12-08
[00:49] nekoluce: hello
[01:01] nekoluce: i didnt saw your messages they werent uploading in the website
[01:02] nekoluce: sorry
[01:02] nekoluce: did you receive my email?
[01:11] nekoluce: hope you had a productive day
[01:12] nekoluce: and you sleep well
[01:51] jfw: I'm a bit confused, nekoluce: if the messages didn't show on the website then how did you see them now?
[01:55] jfw: I did see your email about the ISD event, yes. I haven't had the chance to look up all these things yet but it's good that you mention them.
[02:02] jfw: English point I've seen come up before: "saw" is indeed the past tense of "see", but when you use "do" as a helper verb, only the "do" is conjugated and the main verb takes the infinitive. so it's "I didn't see".
[02:03] jfw: technically you can say the negative in the simple past, "I saw not", but it's out of fashion and you'll sound like Shakespeare.
[02:19] jfw: oh - perhaps you meant they weren't updating on the website? like you saw them only after reloading? that would be normal, as we aren't scripting anything in the browser to poll for updates
[02:22] jfw: and no worries, you don't have to check in constantly; but we did perhaps miss [infinitive, see?] a chance to converse
[12:09] nekoluce: I kept refreshing but they werent loading on my side that is what i meant but good morning.
[12:10] nekoluce: hope you have a good day today as well.
[18:46] jfw: thank you nekoluce. I haven't noticed that problem with the messages but you can show me if it happens again.
[19:26] nekoluce: you are welcome
Day changed to 2024-12-09
[18:10] jfw: it finally happened: an unannounced water cut right in the middle of my shower. just rinsed my hair with spring water from the jug at $1.25/gal (330 sat/L for the standard units). panama life !
[18:18] jfw: I get it though, cutting things off is the only form of communication people understand, since language was replaced by emoji
Day changed to 2024-12-10
[04:17] lru: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12888 - is that 2.5x based on bitcoin going up in price on its own, or was there some additional investment strategy that is not clear from the summary?
[04:17] sourcerer: 2024-12-07 21:52:46 (#jwrd) jfw: someone taking jwrd's advice at that time is up 2.5x today or about 38% yearly without leverage.
[17:55] jfw: lru, it's based on the bitcoin price along with doing the things needed to safely hold, transact & trade in it.
[17:58] jfw: though as far as leverage, there was an allusion to the UPW.
[19:40] lru: jfw: thanks
[20:46] jfw: yw
Day changed to 2024-12-11
[14:18] nekoluce: Bonjour cava bien ?
[14:47] nekoluce: hello xissburg !
Day changed to 2024-12-18
[16:16] jfw: dorion: just looked over the updated group pricing models for training. since it seems to keep running into this perverse incentive against adding more people, how about doing it like the bitcoin supply? say for one student the price is X. then the second adds X/2, though the total is split so each pays 3/4 X (the same as your current plan, as it happens). the next adds X/4, then X/8 etc. Thus,
[16:16] jfw: each one is always adding something to the gross. Could do an inverse instead of geometric series if it falls off too fast (X + X/2 + X/3 + X/4...), and otherwise tweak parameters as desired.
[16:20] jfw: I'll sketch it up in a spreadsheet
[16:28] jfw: http://jfxpt.com/paste/rx563fyqv2 for geometric, asymptotically approaches 2X total gross
[16:31] jfw: http://jfxpt.com/paste/ypepgvvv4s for inverse; total gross grows logarithmically. seems more sensible.
Day changed to 2024-12-20
[20:09] mechs: hi guys, its been almost a year
Day changed to 2024-12-21
[01:25] dorion: welcome back mechs, what's new ?
Day changed to 2024-12-22
[01:43] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12761 - the drives are in hand; still waiting on the cabling.
[01:43] sourcerer: 2024-12-02 04:55:25 (#jwrd) jfw: I went with a slightly updated drive model, ST4000NM0025 which supports the new 12 Gbit SAS link speed, advertises slightly better reliability numbers, higher sustainable throughput and lower power draw, and maintains 512N sector format.
[01:53] jfw: oh hi mechs, looks like it's been since http://jfxpt.com/2023/jwrd-logs-for-Oct-2023/#9861
[01:53] sourcerer: 2023-10-30 20:14:09 (#jwrd) mechs: was this the norm before 2004?
[01:53] jfw: how has the year treated you, or you it?
[02:01] jfw: and since I'm making the rounds, the new node is at 450350 blocks or a dreadful .94 bpm. largely unattended in the interval so probably lots of dead spells.
[02:01] sourcerer: 2024-11-25 17:42:51 (#jwrd) jfw: http://jfxpt.com/2024/jwrd-logs-for-Nov-2024/#12657 - update: 415012 = 1.8 bpm since then.
[05:40] mechs: its been a great year, much to learn still. Instead of doing SICP i ended up doing HtDP, by Felleisen et al. It's a great refresher especially on recursion
[22:22] jfw: popcorn time? https://www.reuters.com/world/americas/trump-says-he-might-demand-panama-hand-over-canal-2024-12-22/
[22:33] jfw: mechs: looking into that briefly, I'm curious what those supposedly onerous 'domain knowledge' requirements from SICP were. I never worked through the whole thing but it seemed pretty well self-contained
[22:34] jfw: and the idea that programs can be 'designed' by following some nice orderly process sounds... hm, enticing but dubious
[22:37] jfw: but perhaps it could provide some helpful mental structure at least to start with
[23:44] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12931 - tracing to the source as best I can, Trump wrote similar remarks yesterday though the press didn't pick it up until today's speech: https://trumpstruth.org/statuses/28704 https://trumpstruth.org/statuses/28705
[23:44] sourcerer: 2024-12-22 22:22:56 (#jwrd) jfw: popcorn time? https://www.reuters.com/world/americas/trump-says-he-might-demand-panama-hand-over-canal-2024-12-22/
Day changed to 2024-12-23
[00:02] jfw: dorion, I recall you sharing a letter from one of the major shipping companies some months ago expressing their gravest concerns about the canal's capacity & pricing troubles, do you have that handy for the background?
[00:26] jfw: imagine reversing the canal zone handover after all this time, it'd be quite the blockchain fork. bye bye ciudad del saber, panama pacifico, ancon... not that I think it's likely to come to that, since panama routinely caves in to much lesser pressures.
[14:00] dorion: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12931 -- mulino seems to be in a double bind. panama can't do much besides talk to assert its sovereignty (all the banks have all the money and other paper assets in the us) and if he does give in, there will be riots. the positioned in weakened by the fact panama has given up sovereignty at ever step : from bearer shares, to crs, to giving up the
[14:00] sourcerer: 2024-12-22 22:22:56 (#jwrd) jfw: popcorn time? https://www.reuters.com/world/americas/trump-says-he-might-demand-panama-hand-over-canal-2024-12-22/
[14:00] dorion: fishing rights (used to be 50 miles off the coast, varela cut it down to 12), to covid derpage. they fucked up the copper mine since the 90s, etc.
[15:16] jfw: maybe he can cut the US navy a special transit deal so trump gets a symbolic win?
[15:19] jfw: the rising costs of the canal also look like the result of long-term mismanagement. though through the bitcoin lens - what rising costs? maybe it's the USG that can no longer afford to move a navy
[15:39] jfw: USG prints up a bunch of dollars, then expects its foreign partners to accept those funny dollars for real services at the same old rates. Does it expect to pay its own servicemen in the same manner?
[15:43] jfw: but I suppose it could work, since nobody else has strength or choices left either, and no place to defect to.
[16:27] dorion: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12944 -- continental dollars, greenback, been the m.o. more times than paying with sound money has. at least as far as paying soldiers.
[16:27] sourcerer: 2024-12-23 15:39:51 (#jwrd) jfw: USG prints up a bunch of dollars, then expects its foreign partners to accept those funny dollars for real services at the same old rates. Does it expect to pay its own servicemen in the same manner?
[16:30] dorion: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12943 -- the braindamaged 'design' of using tug boats on the expanded canal instead of the time tested rail mules has cost A LOT. way more accidents/ship damage there. hell, even the innaugural chicom ship that went through it rubbed up.
[16:30] sourcerer: 2024-12-23 15:19:17 (#jwrd) jfw: the rising costs of the canal also look like the result of long-term mismanagement. though through the bitcoin lens - what rising costs? maybe it's the USG that can no longer afford to move a navy
[16:32] jfw: that and not developing water resources and alternatives to hydro power
[16:33] jfw: any numbers on the collision costs?
Day changed to 2024-12-24
[14:35] nekoluce: Merry christmas
[17:31] jfw: gracias nekoluce, feliz navidad.
[17:32] jfw: I'm finally an officially licensed panamanian driver - those mean streets had better watch out!
[17:34] jfw: they have simulators now to do the practical test and I can certify that they're just as scammy as the rest of it
[17:38] jfw: they're... sorta realistic I guess but still felt much more like a video game than actually operating heavy machinery. so it doesn't really prove you can sense the vehicle, maintain control at a range of speeds or even park worth a damn; but they get to throw up obstacles/hazards that they couldn't in real life
[17:39] jfw: I was going for both car and motorcycle and the car was pretty easy to pass
[17:40] jfw: the hardware on the motorcycle was pretty realistic but the test was configured only in 'scooter' mode with no shifting or anything, and the control/feedback was just awful, almost binary like pressing left/right arrows and yet sensitive, so tough to stay on the intended line
[17:42] jfw: I got a touch of motion sickness, and it failed me on the test for unknown reasons. the guy let me try once more, the second time I got 3 instead of 8 'safety errors' but still made no sense. like I didn't work the blinkers within the designated range of timings or something
[17:45] jfw: you were supposed to 'share the road with a cyclist' and pass him safely, while a car pops into being out of nowhere in the opposing lane. so possibly passing him too close in the right lane fails you and passing in the left lane fails you if there's something wrong with the timing, like maybe there's a pixellated intersection on the left that i didn't spot or something
[17:47] jfw: all he could offer me was to repeat the nonsense on an appointment for another day, so I asked to downgrade the license request to cars only, then had to wait a while longer wondering if I was even still in the queue.
[17:49] jfw: but all good eventually; the practice booklet prepared me well for the exam-taking part which is based on totally obscure details from the local regulations.
[17:54] jfw: also the drive cable & connector options are in hand - slightly ahead of prediction.
[17:54] sourcerer: 2024-12-02 00:52:21 (#jwrd) jfw: there's enough clearance between drives that the right-angle MIGHT work, and enough to the back case wall that the straight connectors MIGHT fit with some gentle cable pre-bending. so it seems clear enough to go ahead with SAS, but will have to grab 'one of each' on the cabling and see what works.
[17:54] sourcerer: 2024-12-02 04:52:26 (#jwrd) jfw: order placed - most of it should be in miami by next tuesday the 10th. by which point I expect the forwarders will be in holiday backlog and maybe it arrives by new year's.
[21:06] nekoluce: thank you for saying merry christmas and im glad you have your license.
[22:39] nekoluce: hope you guys enjoy your christmas i got a new haircut
Day changed to 2024-12-28
[16:41] jfw: bad news on the drives for the server - while billed as new, on closer visual inspection they've clearly had some rough handling, and not just the sort from damage in shipment either but prior use and possibly even attempts to refurbish/hide it.
[16:43] jfw: the right-angle SAS to SATA adapters don't work because the plastic on the side extends a bit too far to fit through the metal cutouts on this 'cable managed' case.
[16:45] jfw: but the SAS fanout cable with molex power feeds works in place of the prior SATA fanout cable connecting to the 3ware RAID controller, after some annoyance with routing all the cables ('managing') through the slim space under the right side panel
[16:48] jfw: the controller sees the drives and maps a volume successfully. the Linux system on the old drives no longer boots, presumably due to the drives no longer enumerating in the same order, and the prior rescue environment didn't include the software RAID utility so I'll need to return to sort that out.
[16:49] jfw: but the rescue environment did provide enough for me to check a hexdump of the new volumes, which confirmed the used condition because there was some data on the first one (with strings involving EFI)
[17:00] jfw: the plan is to raise it with the seller and see if they'll give a refund, meanwhile we can continue testing.
[17:19] jfw: the downside of switching to the SAS cable is we lose the option of also connecting SATA drives through the RAID controller. they can still connect directly to the motherboard though.
[17:20] jfw: possibly I'll rig up something involving UUIDs to harden the Linux RAID configuration against changes in drive order. even without udev, there's at least a 'blkid' command in e2fsprogs
[22:52] nekoluce: heyp
Day changed to 2024-12-29
[01:15] jfw: ohayo nekoluce
[01:20] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12971 - the server's back online, still using hard-wired device node paths for 'mdadm --assemble' in the initramfs boot script. it turns out UUIDs don't actually work despite all the noise, who could've predicted.
[01:20] sourcerer: 2024-12-28 16:48:14 (#jwrd) jfw: the controller sees the drives and maps a volume successfully. the Linux system on the old drives no longer boots, presumably due to the drives no longer enumerating in the same order, and the prior rescue environment didn't include the software RAID utility so I'll need to return to sort that out.
[01:24] jfw: specifically: blkid and the simpler findfs use identifers in the ext* filesystem headers, but that's above the RAID and block device levels so useless for identifying the disks as a whole, and while mdadm can create a UUID of some sort for the RAID volume, it apparently doesn't for the underlying devices and so can't find them by ID.
[01:42] jfw: to fully illustrate the point about "useless for identifying the disks as a whole", "blkid" lists *the same UUID* for the two partitions underlying a RAID1 with mdadm's 0.90 metadata format (which was required for LILO booting), and no UUIDs at all for the partitions underlying a RAID1 with the default metadata format.
[12:23] nekoluce: ohayo
[12:26] nekoluce: I respond your email
[12:26] nekoluce: nice to be back in the chat .w.
[21:08] nekoluce: Using `blkid` to identify RAID1 disks is like asking identical twins wearing masks to tell you their names—either they give you the same answer or no answer at all!
[22:06] jfw: so it seems, yes. silly me, expecting the twins to have names of their own and everything
[23:04] nekoluce: heeeey
Day changed to 2024-12-30
[21:30] nekoluce: good afternoon
[21:50] jfw: hallo
Day changed to 2024-12-31
[23:41] jfw: nekoluce: to decode this response a bit: your bot-assisted analogy has some validity and perhaps some value to help understand the situation, since a RAID1 means replicating identical copies of data across two or more disks. Nonetheless, the attempted humor didn't quite hit the mark for me, because as I point out, even those twins each have
[23:41] sourcerer: 2024-12-29 22:06:45 (#jwrd) jfw: so it seems, yes. silly me, expecting the twins to have names of their own and everything
[23:41] jfw: their own identity despite looking the same at first glance; just so with the disks, there's no reason in principle they couldn't have their own unique identifiers (even if 'blkid' isn't the right thing to check them) and my use case shows how that might be helpful. Basically, I maintain the position that my expectations were reasonable and the avaiable tools fell short, not the other way around.
[23:42] jfw: I wasn't hurt though and you're welcome to poke fun at me, just know that I might poke back if it's warranted!
[02:02] jfw: it looks like it's indeed a much more recent shift on the enterprise HDD side, with datasheets for older models indicating 512-byte sectors (with 520 or 528 byte formatting options) without mention of some larger underlying; this includes the 4TB 'Constellation ES.3' set we got before; more current seagate products in lower capacities show '512N SAS' as a format option, suggesting 'native' as
[02:02] jfw: opposed to the new 512e term. but the peak there seems to be Exos 7E10 6TB, which isn't showing up for sale on their own store or the usual platforms.
[02:28] jfw: yet another speedbump is the cabling: the breakout 'squid' cable from the 3ware controller terminates in SATA type connectors (with the L-shaped edge keying), which accomodates the supermicro backplane which passes it through to the full SAS drive connection including power
[02:30] jfw: but for individual SAS drives in a workstation without the backplane, it seems an adapter is needed because neither SATA data nor power connectors will fit directly onto the SAS drive.
[02:33] jfw: and for added excitement, with this 'cable management' style that all the ATX cases switched to, there's little room for connectors coming out of the drive as they're positioned right against the back panel of the case. would need to check if I can mount them in the opposite direction.
[02:39] jfw: https://www.amazon.com/EDIMS-Internal-SFF-8087-SFF-8482-29Pin/dp/B07HL4M1P2/ looks like the ticket
[03:06] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12736 - for completeness, I did find the nominally 512N variant of that 6T Exos drive with SATA interface (model ST6000NM000B), at CDW.
[03:06] sourcerer: 2024-12-01 02:02:49 (#jwrd) jfw: opposed to the new 512e term. but the peak there seems to be Exos 7E10 6TB, which isn't showing up for sale on their own store or the usual platforms.
[03:10] jfw: so it seems I face an entirely underinformed decision of whether I care most about interface protocol elitism, sector size traditionalism, or raw capacity.
[03:13] jfw: my bias is toward the original plan of ~10T SAS. the sector size thing looks like a lost cause with just this one specific item available today and probably not even in production anymore
[03:20] jfw: western digital's site won't even load a product catalog, with scripts
[03:46] jfw: https://www.sqlite.org/psow.html - some evidence of the sadness of big sectors.
[03:47] jfw: along with https://www.sqlite.org/atomiccommit.html
[03:55] jfw: I don't know of a comparable explanation for MySQL, but there's indication that some operations are not 4k aligned
[06:37] jfw: dorion: new plan: open up the machine tomorrow to check on the question of space for SAS connector (as is or by drive reversibility); if all is well, order 2 more of *the same 4T seagate drive as before* (found it for $90). going all the way on the 'standardize' direction. basically, we know we have database applications, some of which may be write-intensive. we know we need reliability (and have
[06:37] jfw: had various troubles with SATA stuff in the past). whereas we don't as yet have any projects that require that many terabytes; if we take some on, they can better inform the specific hardware needs if indeed there's no longer such thing as the best all-around drive. maybe we'd want something even bigger, or more options come on the market for sector size (one can dream)
[19:48] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12739 - just from checking the manual illustrations I'm recalling the problem at least with this case: it uses slide-out plastic caddies to mount the 3.5" drives, and even if it's physically possible to attach the drive with ports facing the front, the cover plate of the caddy will block the cables.
[19:48] sourcerer: 2024-12-01 02:33:57 (#jwrd) jfw: and for added excitement, with this 'cable management' style that all the ATX cases switched to, there's little room for connectors coming out of the drive as they're positioned right against the back panel of the case. would need to check if I can mount them in the opposite direction.
[19:52] jfw: the only things I'm turning up with right-angled connectors on the SAS drive end (SFF-8482) are right-angle adapters that turn it into SATA type connectors, which is annoying but might work - use the current breakout cable with its SATA connectors, then molex to SATA power adapters as necessary.
[19:55] jfw: the prior linked breakout cable would have done exactly the desired thing without further adapting... but what can you do.
[20:03] jfw: but the right angle could create its own problems with fitting the drives next to each other. so I guess it's still worth checking if those caddies can be sawed open or something.
[20:16] jfw: altogether, computer assembly is starting to look a lot like contemporary software development: plenty of feature-filled frameworks around which economy compels you to adopt and then you spend all your time tracking down their bugs or working around their spurious limitations.
Day changed to 2024-12-02
[00:49] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12755 - so, the only way a drive will go reversed into the caddy is by fully removing at least the clips that snap the caddy into the chassis. there aren't enough screw holes to mount the drive 'manually' without the caddy, but it looks like there's at least one usable to secure it from sliding if the clips are removed.
[00:49] sourcerer: 2024-12-01 20:03:36 (#jwrd) jfw: but the right angle could create its own problems with fitting the drives next to each other. so I guess it's still worth checking if those caddies can be sawed open or something.
[00:52] jfw: there's enough clearance between drives that the right-angle MIGHT work, and enough to the back case wall that the straight connectors MIGHT fit with some gentle cable pre-bending. so it seems clear enough to go ahead with SAS, but will have to grab 'one of each' on the cabling and see what works.
[04:52] jfw: order placed - most of it should be in miami by next tuesday the 10th. by which point I expect the forwarders will be in holiday backlog and maybe it arrives by new year's.
[04:55] jfw: I went with a slightly updated drive model, ST4000NM0025 which supports the new 12 Gbit SAS link speed, advertises slightly better reliability numbers, higher sustainable throughput and lower power draw, and maintains 512N sector format.
[13:54] dorion: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12756 -- pretty sad, but thanks for wading through the swamps.
[13:54] sourcerer: 2024-12-01 20:16:49 (#jwrd) jfw: altogether, computer assembly is starting to look a lot like contemporary software development: plenty of feature-filled frameworks around which economy compels you to adopt and then you spend all your time tracking down their bugs or working around their spurious limitations.
[13:54] dorion: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12759-- sounds good, I'll keep an eye out for the package.
[13:54] sourcerer: 2024-12-02 00:52:21 (#jwrd) jfw: there's enough clearance between drives that the right-angle MIGHT work, and enough to the back case wall that the straight connectors MIGHT fit with some gentle cable pre-bending. so it seems clear enough to go ahead with SAS, but will have to grab 'one of each' on the cabling and see what works.
Day changed to 2024-12-04
[16:57] nekoluce: hello
[17:46] dorion: hello nekoluce, welcome. who might you be ?
[18:22] jfw: Luz found her way in.
[18:26] jfw: nekoluce: I remembered the Android IRC client I had working once, it's Atomic.
[18:26] jfw: now we just need that bouncer setup...
[18:29] jfw: dorion: I was thinking of posting a bounty for that, not because it's hard but because it's relatively self-contained, easy to specify and verify, in the spirit of offering ways in
[18:31] jfw: basically, take your pick of Gales or CentOS, pick the software, version & dependencies, write the recipe to build from source, install, configure, and create user IDs
[18:33] jfw: maybe .002 BTC or you think that's too rich?
[18:36] jfw: last night's quit/join noise btw was from a planned router maintenance at hetzner.
[18:39] jfw: I need to fix our bots to rejoin properly - wotbot has the code working fine to detect timeouts but it just needs to be run in a service to auto-restart; sourcerer is the awk based prototype and doesn't detect the server's absence; the thought was to implement the logging stuff in the new 'production' framework
[18:40] jfw: wouldn't mind paying to get that done either but it's probably a higher code-learning investment up front
[19:00] jfw: hi nekoluce, here's an example of the use of the log and quoting: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12769
[19:00] sourcerer: 2024-12-04 18:26:40 (#jwrd) jfw: nekoluce: I remembered the Android IRC client I had working once, it's Atomic.
[19:03] nekoluce: "Quoting" ?
[19:03] jfw: yes, see how I just pasted the link and the bot replied with the referenced line
[19:04] nekoluce: i'm trying to click on the link but it doesnt seem to work .
[19:05] jfw: maybe right clicking?
[19:07] nekoluce: it appears as the same website ?
[19:08] nekoluce: is it normal ?
[19:08] jfw: it should highlight and scroll to the specific line
[19:08] nekoluce: it is highlighted on the website but not on lime
[19:09] nekoluce: did you arrive safe ?
[19:10] jfw: barely; I was attacked by a row of hungry ducklings.
[19:11] nekoluce: you know i do have a park of hungry ducklings nearby my home jaja.
[19:11] nekoluce: there is even a lake
[19:12] jfw: nice, we usually get some families near my parents' place in Maine
[19:13] nekoluce: What is Maine like ? Cold or warm ?
[19:14] jfw: seasonal; it's pretty far north, on a level with southern Canada, so long winters but temperatures moderated by the sea. summers are historically quite pleasant but last few years have been hot.
[19:15] nekoluce: hot as panama ? it seems beautiful and lovely scenery .
[19:15] jfw: it can be, but not as humid.
[19:15] nekoluce: wish someday I meet Maine owo
[19:16] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12786 - that makes sense, it's just a feature of the web view. but that's why the bot quoting helps to follow right from the client.
[19:16] sourcerer: 2024-12-04 19:08:50 (#jwrd) nekoluce: it is highlighted on the website but not on lime
[19:16] nekoluce: hi sourcerer !
[19:25] nekoluce: it has a password for atomic
[19:26] nekoluce: sorry for not saying hi rob
[19:49] jfw: like a password on startup, really? odd since it's supposed to be open source. did you grab it from play store or f-droid?
[19:53] jfw: I have version 2.1
[20:08] nekoluce: from fdroid
[20:17] nekoluce: oh it work
[20:17] nekoluce: but i have another username
[20:17] nekoluce: so I have now two ?
Day changed to 2024-12-05
[03:48] jfw: the fiat world hits a new benchmark low with one USD now fetching less than 1000 satoshi.
[03:51] lru: and I finally bought some
[03:52] jfw: I guess it's time for the correction then!
[03:52] lru: lol
[03:52] lru: sorry everyone :-)
[03:52] lru: j/k
[03:53] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12807 - an irc nickname can only be held by one connection at a time. so you can set the second client to use a variant of the name, or just use one at a time. I think this would also be improved with a bouncer setup
[03:53] sourcerer: 2024-12-04 20:17:44 (#jwrd) nekoluce: so I have now two ?
[03:53] jfw: well congrats on dipping the next toe in, lru.
[03:54] lru: thanks
[03:55] lru: how do your bounties work btw? is it first person who completes gets paid, or first person who accepts the challenge and then completes in reasonable time?
[03:57] jfw: tbd, but in this case I'd be inclined to the latter
[03:58] jfw: i.e. it's a commitment and what you make of it will reflect one way or another
[03:59] lru: sounds good... I suppose if there is a crowd wanting to participate, nothing stopping people splitting the reward and working together
[04:00] jfw: WoT identity is a must, needless to say. and sure, it might be delegated but someone in particular will be responsible
[04:06] lru: I see
[04:06] lru searches for wotbot docs
[04:10] lru: found 'em
[11:17] Atomic_0TTFW: good morning 🌅
[11:18] Atomic_0TTFW: ohayogozaimasu ! desu
[14:46] nekoluce: hello mr xissburg
[15:09] xissburg: nekoluce: neko ga imasu!
[15:10] nekoluce: how is your day ?
[15:10] xissburg: glorious day
[15:11] nekoluce: amazing having a nice weather.w. ?
[15:14] nekoluce: i started using irc yesterday is quite fun
[20:20] jfw: what's this, secret codes? unfair!!
[20:25] jfw: I'm getting "good morning! it is" and "there is a cat"
[20:42] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12773 - I picked that number because it's about what I'd want if doing it for someone else; someone less experienced or familiar with the environment would take longer and get less per hour, so that seems fine; and if it's "too much" then it could still be a good idea for now, viewing the excess as publicity expense. if at some point we end up with
[20:42] sourcerer: 2024-12-04 18:33:04 (#jwrd) jfw: maybe .002 BTC or you think that's too rich?
[20:42] jfw: more qualified applicants than we know what to do with, then we can look at competitive bidding
[20:46] jfw: xissburg, you got any war stories from hiring engineers or gig workers generally?
[20:50] nekoluce: hey are you guys going to the btc party ?
[20:50] xissburg: jfw: hm not really, just firing due to incompetence
[20:50] xissburg: do you?
[20:51] xissburg: nekoluce: nah
[20:52] jfw: xissburg: just dipping my toes in now
[20:52] nekoluce: having a bath ?
[20:52] jfw: nekoluce: heh, not literally
[20:53] nekoluce: ah xD
[20:53] jfw: what's the btc party you mention?
[20:54] jfw: the offchain guys doing something?
[20:54] nekoluce: tonight there is a btc party at
[20:54] nekoluce: yes the offchain guys
[20:54] nekoluce: like three groups together
[20:55] nekoluce: the biergarten
[20:55] nekoluce: since 6
[20:55] nekoluce: pm
[20:55] jfw: perhaps I'll stop by
[20:55] nekoluce: yay. !
[20:56] xissburg: jfw: I think it's good to be ready to fail fast/fire fast
[20:58] jfw: right, basically you make yourself vulnerable to some extent so see what people do with that and don't extend more credit than necessary
[20:59] nekoluce: thank you xissburg ,bye bye Ill go get ready
[21:02] jfw: xissburg: any particular venues you've found good for announcing offers, or do you find the word spreads itself if the offer is good?
[21:17] xissburg: jfw: you have to announce it where the likely candidates hang out
Day changed to 2024-12-06
[16:40] nekoluce: Good morning !
[18:27] jfw: good day nekoluce
[18:28] nekoluce: is it rainy at your house ?
[18:30] jfw: yes, a light rain
[18:32] nekoluce: rain is lovely makes panama less hot and more cloudy those days are fun with hot cocoa .
[18:33] nekoluce: your welcome
[18:34] jfw: I find hot cocoa improves just about any day indeed.
[18:34] jfw: and I like the rains here except for those times when I'm on foot and the roads are flooding.
[18:35] nekoluce: oh ,yes .
[18:35] jfw: just a different kind of adventure though I guess.
[18:36] nekoluce: btw there is a guy in one of the groups promoting an event maybe it could help get exposure for you guys toward more people in the market is called Panama Future .I dont know if you are interested ?
[18:38] jfw: couldn't hurt to have a look, does it have a site/article/anything?
[18:38] nekoluce: actuaally there are two
[18:38] nekoluce: Panama Blockchain week also is another one
[18:38] nekoluce: let me look for it
[18:41] jfw: xissburg: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12862 - makes sense, so figuring that out is part of the challenge
[18:41] sourcerer: 2024-12-05 21:17:57 (#jwrd) xissburg: jfw: you have to announce it where the likely candidates hang out
Day changed to 2024-12-07
[16:37] nekoluce: hello
[21:31] jfw: hi nekoluce
[21:32] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12873 - I guess that would be https://panamafuture.com/
[21:32] sourcerer: 2024-12-06 18:36:53 (#jwrd) nekoluce: btw there is a guy in one of the groups promoting an event maybe it could help get exposure for you guys toward more people in the market is called Panama Future .I dont know if you are interested ?
[21:34] jfw: I'd see that as scammy until proven innocent. blockchain tokens for real estate? property rights in real estate are ultimately arbitered by the ruling powers on the ground, while the block chain exists outside of any particular state. those mix like oil and water.
[21:35] jfw: serve up some marketing word salad and it'll provide the dressing, sure.
[21:47] jfw: this might mark a key difference between us and perhaps noisier crypto bros. at that party the other night they went around sharing, almost boasting it seemed, about their biggest losses, how they've been robbed but keep getting back up. taking risks is necessary to get anywhere, sure - but get educated and take smart risks.
[21:47] jfw: "don't get robbed" is pretty central to my recipe for wealth, as I talked about for instance here.
[21:52] jfw: someone taking jwrd's advice at that time is up 2.5x today or about 38% yearly without leverage.
[21:55] jfw: in dollar terms, sure, for the sake of comparing investments.
Day changed to 2024-12-08
[00:49] nekoluce: hello
[01:01] nekoluce: i didnt saw your messages they werent uploading in the website
[01:02] nekoluce: sorry
[01:02] nekoluce: did you receive my email?
[01:11] nekoluce: hope you had a productive day
[01:12] nekoluce: and you sleep well
[01:51] jfw: I'm a bit confused, nekoluce: if the messages didn't show on the website then how did you see them now?
[01:55] jfw: I did see your email about the ISD event, yes. I haven't had the chance to look up all these things yet but it's good that you mention them.
[02:02] jfw: English point I've seen come up before: "saw" is indeed the past tense of "see", but when you use "do" as a helper verb, only the "do" is conjugated and the main verb takes the infinitive. so it's "I didn't see".
[02:03] jfw: technically you can say the negative in the simple past, "I saw not", but it's out of fashion and you'll sound like Shakespeare.
[02:19] jfw: oh - perhaps you meant they weren't updating on the website? like you saw them only after reloading? that would be normal, as we aren't scripting anything in the browser to poll for updates
[02:22] jfw: and no worries, you don't have to check in constantly; but we did perhaps miss [infinitive, see?] a chance to converse
[12:09] nekoluce: I kept refreshing but they werent loading on my side that is what i meant but good morning.
[12:10] nekoluce: hope you have a good day today as well.
[18:46] jfw: thank you nekoluce. I haven't noticed that problem with the messages but you can show me if it happens again.
[19:26] nekoluce: you are welcome
Day changed to 2024-12-09
[18:10] jfw: it finally happened: an unannounced water cut right in the middle of my shower. just rinsed my hair with spring water from the jug at $1.25/gal (330 sat/L for the standard units). panama life !
[18:18] jfw: I get it though, cutting things off is the only form of communication people understand, since language was replaced by emoji
Day changed to 2024-12-10
[04:17] lru: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12888 - is that 2.5x based on bitcoin going up in price on its own, or was there some additional investment strategy that is not clear from the summary?
[04:17] sourcerer: 2024-12-07 21:52:46 (#jwrd) jfw: someone taking jwrd's advice at that time is up 2.5x today or about 38% yearly without leverage.
[17:55] jfw: lru, it's based on the bitcoin price along with doing the things needed to safely hold, transact & trade in it.
[17:58] jfw: though as far as leverage, there was an allusion to the UPW.
[19:40] lru: jfw: thanks
[20:46] jfw: yw
Day changed to 2024-12-11
[14:18] nekoluce: Bonjour cava bien ?
[14:47] nekoluce: hello xissburg !
Day changed to 2024-12-18
[16:16] jfw: dorion: just looked over the updated group pricing models for training. since it seems to keep running into this perverse incentive against adding more people, how about doing it like the bitcoin supply? say for one student the price is X. then the second adds X/2, though the total is split so each pays 3/4 X (the same as your current plan, as it happens). the next adds X/4, then X/8 etc. Thus,
[16:16] jfw: each one is always adding something to the gross. Could do an inverse instead of geometric series if it falls off too fast (X + X/2 + X/3 + X/4...), and otherwise tweak parameters as desired.
[16:20] jfw: I'll sketch it up in a spreadsheet
[16:28] jfw: http://jfxpt.com/paste/rx563fyqv2 for geometric, asymptotically approaches 2X total gross
[16:31] jfw: http://jfxpt.com/paste/ypepgvvv4s for inverse; total gross grows logarithmically. seems more sensible.
Day changed to 2024-12-20
[20:09] mechs: hi guys, its been almost a year
Day changed to 2024-12-21
[01:25] dorion: welcome back mechs, what's new ?
Day changed to 2024-12-22
[01:43] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12761 - the drives are in hand; still waiting on the cabling.
[01:43] sourcerer: 2024-12-02 04:55:25 (#jwrd) jfw: I went with a slightly updated drive model, ST4000NM0025 which supports the new 12 Gbit SAS link speed, advertises slightly better reliability numbers, higher sustainable throughput and lower power draw, and maintains 512N sector format.
[01:53] jfw: oh hi mechs, looks like it's been since http://jfxpt.com/2023/jwrd-logs-for-Oct-2023/#9861
[01:53] sourcerer: 2023-10-30 20:14:09 (#jwrd) mechs: was this the norm before 2004?
[01:53] jfw: how has the year treated you, or you it?
[02:01] jfw: and since I'm making the rounds, the new node is at 450350 blocks or a dreadful .94 bpm. largely unattended in the interval so probably lots of dead spells.
[02:01] sourcerer: 2024-11-25 17:42:51 (#jwrd) jfw: http://jfxpt.com/2024/jwrd-logs-for-Nov-2024/#12657 - update: 415012 = 1.8 bpm since then.
[05:40] mechs: its been a great year, much to learn still. Instead of doing SICP i ended up doing HtDP, by Felleisen et al. It's a great refresher especially on recursion
[22:22] jfw: popcorn time? https://www.reuters.com/world/americas/trump-says-he-might-demand-panama-hand-over-canal-2024-12-22/
[22:33] jfw: mechs: looking into that briefly, I'm curious what those supposedly onerous 'domain knowledge' requirements from SICP were. I never worked through the whole thing but it seemed pretty well self-contained
[22:34] jfw: and the idea that programs can be 'designed' by following some nice orderly process sounds... hm, enticing but dubious
[22:37] jfw: but perhaps it could provide some helpful mental structure at least to start with
[23:44] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12931 - tracing to the source as best I can, Trump wrote similar remarks yesterday though the press didn't pick it up until today's speech: https://trumpstruth.org/statuses/28704 https://trumpstruth.org/statuses/28705
[23:44] sourcerer: 2024-12-22 22:22:56 (#jwrd) jfw: popcorn time? https://www.reuters.com/world/americas/trump-says-he-might-demand-panama-hand-over-canal-2024-12-22/
Day changed to 2024-12-23
[00:02] jfw: dorion, I recall you sharing a letter from one of the major shipping companies some months ago expressing their gravest concerns about the canal's capacity & pricing troubles, do you have that handy for the background?
[00:26] jfw: imagine reversing the canal zone handover after all this time, it'd be quite the blockchain fork. bye bye ciudad del saber, panama pacifico, ancon... not that I think it's likely to come to that, since panama routinely caves in to much lesser pressures.
[14:00] dorion: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12931 -- mulino seems to be in a double bind. panama can't do much besides talk to assert its sovereignty (all the banks have all the money and other paper assets in the us) and if he does give in, there will be riots. the positioned in weakened by the fact panama has given up sovereignty at ever step : from bearer shares, to crs, to giving up the
[14:00] sourcerer: 2024-12-22 22:22:56 (#jwrd) jfw: popcorn time? https://www.reuters.com/world/americas/trump-says-he-might-demand-panama-hand-over-canal-2024-12-22/
[14:00] dorion: fishing rights (used to be 50 miles off the coast, varela cut it down to 12), to covid derpage. they fucked up the copper mine since the 90s, etc.
[15:16] jfw: maybe he can cut the US navy a special transit deal so trump gets a symbolic win?
[15:19] jfw: the rising costs of the canal also look like the result of long-term mismanagement. though through the bitcoin lens - what rising costs? maybe it's the USG that can no longer afford to move a navy
[15:39] jfw: USG prints up a bunch of dollars, then expects its foreign partners to accept those funny dollars for real services at the same old rates. Does it expect to pay its own servicemen in the same manner?
[15:43] jfw: but I suppose it could work, since nobody else has strength or choices left either, and no place to defect to.
[16:27] dorion: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12944 -- continental dollars, greenback, been the m.o. more times than paying with sound money has. at least as far as paying soldiers.
[16:27] sourcerer: 2024-12-23 15:39:51 (#jwrd) jfw: USG prints up a bunch of dollars, then expects its foreign partners to accept those funny dollars for real services at the same old rates. Does it expect to pay its own servicemen in the same manner?
[16:30] dorion: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12943 -- the braindamaged 'design' of using tug boats on the expanded canal instead of the time tested rail mules has cost A LOT. way more accidents/ship damage there. hell, even the innaugural chicom ship that went through it rubbed up.
[16:30] sourcerer: 2024-12-23 15:19:17 (#jwrd) jfw: the rising costs of the canal also look like the result of long-term mismanagement. though through the bitcoin lens - what rising costs? maybe it's the USG that can no longer afford to move a navy
[16:32] jfw: that and not developing water resources and alternatives to hydro power
[16:33] jfw: any numbers on the collision costs?
Day changed to 2024-12-24
[14:35] nekoluce: Merry christmas
[17:31] jfw: gracias nekoluce, feliz navidad.
[17:32] jfw: I'm finally an officially licensed panamanian driver - those mean streets had better watch out!
[17:34] jfw: they have simulators now to do the practical test and I can certify that they're just as scammy as the rest of it
[17:38] jfw: they're... sorta realistic I guess but still felt much more like a video game than actually operating heavy machinery. so it doesn't really prove you can sense the vehicle, maintain control at a range of speeds or even park worth a damn; but they get to throw up obstacles/hazards that they couldn't in real life
[17:39] jfw: I was going for both car and motorcycle and the car was pretty easy to pass
[17:40] jfw: the hardware on the motorcycle was pretty realistic but the test was configured only in 'scooter' mode with no shifting or anything, and the control/feedback was just awful, almost binary like pressing left/right arrows and yet sensitive, so tough to stay on the intended line
[17:42] jfw: I got a touch of motion sickness, and it failed me on the test for unknown reasons. the guy let me try once more, the second time I got 3 instead of 8 'safety errors' but still made no sense. like I didn't work the blinkers within the designated range of timings or something
[17:45] jfw: you were supposed to 'share the road with a cyclist' and pass him safely, while a car pops into being out of nowhere in the opposing lane. so possibly passing him too close in the right lane fails you and passing in the left lane fails you if there's something wrong with the timing, like maybe there's a pixellated intersection on the left that i didn't spot or something
[17:47] jfw: all he could offer me was to repeat the nonsense on an appointment for another day, so I asked to downgrade the license request to cars only, then had to wait a while longer wondering if I was even still in the queue.
[17:49] jfw: but all good eventually; the practice booklet prepared me well for the exam-taking part which is based on totally obscure details from the local regulations.
[17:54] jfw: also the drive cable & connector options are in hand - slightly ahead of prediction.
[17:54] sourcerer: 2024-12-02 00:52:21 (#jwrd) jfw: there's enough clearance between drives that the right-angle MIGHT work, and enough to the back case wall that the straight connectors MIGHT fit with some gentle cable pre-bending. so it seems clear enough to go ahead with SAS, but will have to grab 'one of each' on the cabling and see what works.
[17:54] sourcerer: 2024-12-02 04:52:26 (#jwrd) jfw: order placed - most of it should be in miami by next tuesday the 10th. by which point I expect the forwarders will be in holiday backlog and maybe it arrives by new year's.
[21:06] nekoluce: thank you for saying merry christmas and im glad you have your license.
[22:39] nekoluce: hope you guys enjoy your christmas i got a new haircut
Day changed to 2024-12-28
[16:41] jfw: bad news on the drives for the server - while billed as new, on closer visual inspection they've clearly had some rough handling, and not just the sort from damage in shipment either but prior use and possibly even attempts to refurbish/hide it.
[16:43] jfw: the right-angle SAS to SATA adapters don't work because the plastic on the side extends a bit too far to fit through the metal cutouts on this 'cable managed' case.
[16:45] jfw: but the SAS fanout cable with molex power feeds works in place of the prior SATA fanout cable connecting to the 3ware RAID controller, after some annoyance with routing all the cables ('managing') through the slim space under the right side panel
[16:48] jfw: the controller sees the drives and maps a volume successfully. the Linux system on the old drives no longer boots, presumably due to the drives no longer enumerating in the same order, and the prior rescue environment didn't include the software RAID utility so I'll need to return to sort that out.
[16:49] jfw: but the rescue environment did provide enough for me to check a hexdump of the new volumes, which confirmed the used condition because there was some data on the first one (with strings involving EFI)
[17:00] jfw: the plan is to raise it with the seller and see if they'll give a refund, meanwhile we can continue testing.
[17:19] jfw: the downside of switching to the SAS cable is we lose the option of also connecting SATA drives through the RAID controller. they can still connect directly to the motherboard though.
[17:20] jfw: possibly I'll rig up something involving UUIDs to harden the Linux RAID configuration against changes in drive order. even without udev, there's at least a 'blkid' command in e2fsprogs
[22:52] nekoluce: heyp
Day changed to 2024-12-29
[01:15] jfw: ohayo nekoluce
[01:20] jfw: http://jfxpt.com/2024/jwrd-logs-for-Dec-2024/#12971 - the server's back online, still using hard-wired device node paths for 'mdadm --assemble' in the initramfs boot script. it turns out UUIDs don't actually work despite all the noise, who could've predicted.
[01:20] sourcerer: 2024-12-28 16:48:14 (#jwrd) jfw: the controller sees the drives and maps a volume successfully. the Linux system on the old drives no longer boots, presumably due to the drives no longer enumerating in the same order, and the prior rescue environment didn't include the software RAID utility so I'll need to return to sort that out.
[01:24] jfw: specifically: blkid and the simpler findfs use identifers in the ext* filesystem headers, but that's above the RAID and block device levels so useless for identifying the disks as a whole, and while mdadm can create a UUID of some sort for the RAID volume, it apparently doesn't for the underlying devices and so can't find them by ID.
[01:42] jfw: to fully illustrate the point about "useless for identifying the disks as a whole", "blkid" lists *the same UUID* for the two partitions underlying a RAID1 with mdadm's 0.90 metadata format (which was required for LILO booting), and no UUIDs at all for the partitions underlying a RAID1 with the default metadata format.
[12:23] nekoluce: ohayo
[12:26] nekoluce: I respond your email
[12:26] nekoluce: nice to be back in the chat .w.
[21:08] nekoluce: Using `blkid` to identify RAID1 disks is like asking identical twins wearing masks to tell you their names—either they give you the same answer or no answer at all!
[22:06] jfw: so it seems, yes. silly me, expecting the twins to have names of their own and everything
[23:04] nekoluce: heeeey
Day changed to 2024-12-30
[21:30] nekoluce: good afternoon
[21:50] jfw: hallo
Day changed to 2024-12-31
[23:41] jfw: nekoluce: to decode this response a bit: your bot-assisted analogy has some validity and perhaps some value to help understand the situation, since a RAID1 means replicating identical copies of data across two or more disks. Nonetheless, the attempted humor didn't quite hit the mark for me, because as I point out, even those twins each have
[23:41] sourcerer: 2024-12-29 22:06:45 (#jwrd) jfw: so it seems, yes. silly me, expecting the twins to have names of their own and everything
[23:41] jfw: their own identity despite looking the same at first glance; just so with the disks, there's no reason in principle they couldn't have their own unique identifiers (even if 'blkid' isn't the right thing to check them) and my use case shows how that might be helpful. Basically, I maintain the position that my expectations were reasonable and the avaiable tools fell short, not the other way around.
[23:42] jfw: I wasn't hurt though and you're welcome to poke fun at me, just know that I might poke back if it's warranted!
That was funny I'm glad you earn your batch soldier
Comment by Luz Fung — 2024-12-24 @ 17:59
Merry Christmas
Comment by Luz Fung — 2024-12-24 @ 19:53