K‮ly‬e @khird@qoto.org

@louiscouture I've used both heavily at different times. Octave will accept pretty much all valid Matlab code, but the reverse is not true (Matlab doesn't support Octave's use of x++ to increment x, for example, and it's pickier about single vs. double quotes). So if you need to submit code for the grader to run, you should at least do the remote desktop thing afterward to verify your code is Matlab-legal. If you just need to submit your results, Octave's richer syntax is certainly nice to have.

Matlab's user interface beats the pants off Octave. In Matlab I could use the variable window almost like a spreadsheet to edit arrays in-place, delete failed commands from the history window, copy-paste large blocks of commands from the history window into my editor or command line, etc. Octave's mindset is different - the GUI is an afterthought and by default it just runs in terminal.

Matlab has better external tooling. There is a bunch of field-specific "toolboxes" but also a repository of user-submitted functions that you'll only have access to with a Mathworks account, and the licence forbids you from using those functions in anything other than Matlab as I recall. This may be less of a problem now with Github's increased prominence.

@dragfyre a couple thoughts:

First, re: boundaries, an anecdote. A local community member spent some time pioneering in a Pacific island country. He identified some differences in cultural norms, notably that their threshold for eye contact before it was perceived as staring ("bad face") was much lower, and their threshold for personal space was much higher, so that everyone stood several metres apart when queueing up at the McDonald's.
I later imagined the reverse, a visitor from that country to the West - he'd probably feel very uncomfortable that everyone is, from his perspective, staring at him and crowding him, but he has no real standing to demand more separation in the lines at our McDonald's, or that the cashier not look him in the face when serving him. The thought experiment exposed to me a fundamental problem in the way we talk about personal boundaries: we don't, in fact, unilaterally set our own; they're the result of social consensus. The debate over quote toots is a natural and healthy attempt to establish that consensus - a good chunk of the Fediverse now expects that functionality, but many existing instances don't.

Second, the way quotes are implemented here on QOTO is that the text includes a link prefixed with QT. Quote-aware clients simply follow the link and render the toot; the vanilla frontend ignores this and just gives you the raw hyperlink. I don't see any practical way to implement "forbid people elsewhere on the internet to paste a link to my post" with how ActivityPub works, but maybe we don't need to.
My understanding of the anti-quote argument is that it enables unhealthy behaviours because, unlike the reply functionality, it doesn't notify the original author so he can't rebut the quoting author's attached commentary, leading to one-sided "dunking" instead of open discussion. So what if we approached this from the other side? Rather than preventing the behaviour, simply detect it and issue the notification as you would for any other interaction. The technical end would look something like this: when your instance becomes aware of a toot containing a link to a post on your instance, via either ActivityPub or the Referer header of an incoming request, and that link isn't already in-thread wrt the link target, generate a notification for the original author ("@user@example.org quoted your toot") and add the quoted post and its author to the in-reply-to-id and in-reply-to-account-id fields so it'll be treated as in-thread in case you become aware of it again.

@realcaseyrollins if you can cut down the video to a few seconds of dialogue I would think you could upload that to e.g. dropbox under fair use rules. Your commands look pretty much like what I'd be trying though so I think you're on the right track

@realcaseyrollins the documentation suggests it's possible, but I don't think I have anything with PGS subtitles that I could test. If you point me to a test case I can download from somewhere, I'll take a look

@markmccaughrean Miracle Whip tries to position itself this way but I don't think it ever became an idiom like "a bit Marmite".

https://youtube.com/watch?v=hrmIwwQ-sv8
https://youtube.com/watch?v=_-qh11W7Iuo

@realcaseyrollins here's a Bash script that shows what I meant. Working from a random frame is slower than from a timestamp but it gets there in the end. You just have to provide two arguments: video file and subtitle file.

#!/bin/bash

framecount=$(ffprobe -v error -select_streams v:0 -count_packets -show_entries stream=nb_read_packets -of csv=p=0 "$1");
(( randomframe = ( $RANDOM * 2**15 + $RANDOM ) % $framecount ));

ffmpeg -copyts -i "$1" -vf "subtitles=$2,select=gte(n\,$randomframe)" -vframes 1 "$1.png"

@realcaseyrollins looks like you would just have to pass -copyts as an option alongside -vf select

@realcaseyrollins I think a shell script around ffmpeg is the way to go.

-count_frames to get the range
-vf select to just grab a certain frame

@dimsumthinking @Shayman I was trying to make sense of the post and wondered if it meant lunar months. It'd naively be in the 200ish range if they're independent, but, I dunno, maybe there's some weird pattern - a lunar month is pretty close to four weeks, and I wouldn't be shocked to work out the math to find that the excess has lined up just right to make the distribution really nonuniform.

Then I remembered admiring a waxing crescent moon this week and that hypothesis kinda fell apart

@Shayman I made something to that effect actually for the US spec. Very crude but the idea was it superimposes the minimum and maximum heights for various features (chin, eyes, top of head) and you pan/zoom until everything's between the lines, then it crops and positions it so it fits the standard 4"-by-6" size and you can cut it out (which sell for 35c as opposed to 15 or 20 USD to order the 2"-by-2" size required by the passport office).

Obviously it can't check if you're wearing glasses or making a funny face or anything, but with a bit of human judgement it makes it possible to verify there isn't a formatting error that would invalidate it.

The issue for Canadian requirements is that the photos have to be taken by a photographer who signs and dates them, so you'd need to go once, have your photo taken and get a digital copy, verify it, and then go back to have him print, sign, and date the physical version.

This metal plate was just installed recently. Emergencies are now prohibited on the premises.

@Shayman and the dash in a phone number separates the exchange code from the station code (and separates the area code from the exchange code in ten-digit numbers). This means that any town large enough to have its own telephone exchange gets assigned a block of ten thousand numbers (10⁴ station codes) even if it serves far fewer phones than that.

@Shayman

I thought it was gonna be in Edmonton this year actually:

"Holiday Monday features some of the league’s fiercest rivals: the Tiger-Cats welcome the Argos to Hamilton and the Elks host the Stampeders." So it's Stamps fans that'll have to drive.

https://www.cfl.ca/2022/12/13/circle-your-calendars-cfl-releases-full-2023-schedule/

"In Labour Day action, the Ticats host the Argos and the Elks welcome the Stampeders to their stomping grounds at Commonwealth Stadium."

https://www.tsn.ca/cfl/cfl-releases-2023-season-schedule-with-new-twist-1.1892688

@AnthonyReimer

@realcaseyrollins I must be misunderstanding. Generally you can put in a higher-power PSU without ill effect; the notion that a motherboard only supports up to a certain wattage is odd. It draws a certain amount of current at a particular voltage, and your PSU just has to maintain the rail at the correct voltage. A bigger PSU can maintain the voltage in the face of you drawing a greater current, but it isn't going to overvolt your other components. You can even run it completely unloaded and measure the potential across its outputs with a multimeter - it's not going to spike up to infinity trying to force current through the disconnected wires.

There's a partial exception if you swap between consumer/workstation/server/gaming PSUs or ones of very different age. The reason is that there are a few different rails, and, at different points in time and across different market segments, the demand for current on each rail has changed. So you may buy a model that has a higher total rating, but has *less* power available on a particular rail you need, because it's designed for applications where that rail is used less.

@RL_Dane Looking further, I think it was actually hdajackretask, not hda-analyzer