Notices by picofarad (picofarad@noauthority.social), page 4

@BowsacNoodle @SK1ZM @Spingebill change it to "opulent" - weedeater is like "kleenex" - a brand name that got diluted. I'm assuming it's to avoid hiring people that used that prompt injection to trick the AI.

I also don't buy the "20 years of experience in ..." part. but i guess it's worth a shot, although why you'd want to work for a retarded company is beyond me.

@icedquinn I'm so far out of the loop (hrrr) i don't even know what's available, really. Glad you have jerbs

@icedquinn I had an MPC but i got it used and didn't know someone had jammed the CF card in wrong and bent a bunch of pins. I tried to take it apart but that thing is built like a tank, tons of sheetmetal.

ReCycle is a beat slicer https://www.youtube.com/watch?v=AU_8n8k0nWs

@icedquinn it's kinda like propellerhead recycle, then? (used with ReDrum or something, it's been a long time since i've used it)

@sun i skipped around and it appears they just shout different syllables in repeated groups of at least 3 repetitions

How can you tell it was a kirby shirt (the vacuum company, right?)

@icedquinn I was wondering aloud in a text chat the other day about cameras/sensors that can capture the entire sensor at once instead of line by line, and my friend said he has a few of the "three stacked CCD" pro camcorders.

I'm not sure he understood what i was asking about, so i am unsure if merely having no rolling shutter is the same as the full sensor read at the same time

anyhow good luck on your contemplations

@icedquinn I got an older friend that occasionally will post 55 gallon drum pricing for things like personal lubricant with the same sensawonder

@icedquinn my water is "free" except for the marginal cost of electricity, this water is $1.73 per gallon, and walmart (last time i cared to look) had $0.69/gal - definitely less than half the cost regardless.

But what's the issue you take with it? Is it because a tote that size holds >400 gallons and can be filled for some marginal cost (say a quarter ($0.25) a gallon at an RO facility)?

or is it because it's non-returnable, lol

@blitzed i doubt it. It's an election year. This is more pipe dream promises like a "i wanna be freshman class president because i will install soda machines in every hall in the school and also we will have more dances and ice cream will be served in the cafeteria"

https://www.pmcpa.org.uk/media/cwvkqvyz/3741-case-report-28-march-2024.pdf

Pfizer got butt-slammed. I'm shocked, *shocked* that lying was going on

@alarig irc is so dog-simple that after reading about it for a few minutes you can probably open a telnet or netcat session to an ircd and send/receive queries and join channels. so it's not impressive on that part, but a cohesive client in UEFI i guess is impressive. I guess if you have an old PC laying around you could make a "thin" "client" that just "booted" into IRC?

@icedquinn no, but i wrote a small python program that emulates grep output and it runs in 1.17 seconds, which is 0.99 seconds slower than just grep.

anyhow i'm not arguing i figured you'd have some anecdotes or whatever 😺

@icedquinn @icedquinn I am making a lot of assumptions which i'm generally bad at, like "shouldn't both the DB and the file contents be 'hot' if i repeatedly run the commands?" and "python overhead probably accounts for a large portion" but 20ms is gunna be pretty hard to beat

@icedquinn it may, in fact, be able to get the context, but

@icedquinn does the FTS only return that the search term exists in the document/record, and not the part of the record that contains the search term?

I just did an MVP and it only shows the first part of each record, asking a GPT how to get the context (single line) says "here's how to do text search in python"

it's also 50% faster than grepping over CIFS. 4 seconds vs 8.

grepping /tmp/ same files 0.04s.

Guess i'll just use grep 😄

@icedquinn
1 Apache Solr
2 grep
3 sqlite

for a public web interface to 162MB of textual data that will see 5 hits a day median

What do you think?

I've done the grep thing before but the file was only a few MB (a map-reduce of all zip codes to city, state combos and vice versa in the US) along with either maidenhead grid squares or lat/lon, it's been 6 or 7 years since i had it running. That also wasn't a public facing website, it was to respond to a REST call with the weather without a DB

@icedquinn it worked on android and i tracked it down to "the developer of wireguard for ios/macos doesn't care to update the software".

So i told who i was trying to help "tough break, kid"

I want to self-host (in a datacenter) an endpoint that iphone users can use as a "VPN Gateway". I don't want them to have to pay, and i don't want to pay anything above my current hosting costs.

So far openVPN and wireguard have *not* worked. Wireguard won't get internet, and the old "trick" of resetting the VPN connection a few times no longer works.

With openVPN the turnkey appliance didn't work correctly and i don't have the time or desire to set up all the infra ovpn requires.

Raccoon?

@icedquinn slightly off-topic, what's the best voice codec? how many khz / kbit/s? One of these days i'm going to implement something to sound like AM on sideband on HF radio which is ~2.5khz wide

@icedquinn for the last 20 years it's bugged me that there's no real multi-pass compression solution; however there's headway made with ML compression, zip does 1gb wiki dump @ 300MB and current SOTA is 112MB using "AI" whatevers.

If i had to store stuff on tape, say, i'd dedupe and try to force the "already compressed" stuff to the end of the archive. then i'd have to start googling what the SOTA is. I'd also parchive onto a blu-ray just in case.

the wiki thing is allowed 50 hours to unzip!