MrHall

oh, I already wrote a tool that does that.... kidding, looks very nice 😁

huh, you made me read up more carefully and you're right, it's more nuanced than that. it's a bit interesting, they actually share deps where possible. that's genuinely more interesting than I gave it credit for..  I'm not 100% on board over your native packaging but I like it a lot more  I'd definitely use it if distributing, however - but I'd try to have it in AUR and apt too 

I much prefer not having 36 various versions of the same packages installed, most of them unmaintained and not able to be updated by me. what do I know, I just write software for a living 🙄

i hate flatpak, just the idea of shipping every dependency every stupid time, it's as bad as python being unable to manage dependency compatibility to such a degree you now use 'pip' to install a completely different python and deps for each different program. mind you if i was trying to distribute a program for linux, i'd probably suck it up and use flatpak

idk, i use arch and it's pretty easy. yay and pacman make installing/updating very simple

i wouldn't like voice messages, i'm always busy and getting them at a bad time would stress me out food. i would like a nice lunch, food makes me feel very appreciated.

jokes on you, he's saving an awful lot of power right now 

I'm not saying it's not, but I am saying really smart people can do super stupid things when they're hurrying and don't think it's that risky

I'm definitely working, unlike my production environment 😒

this was a giant failure of process, they did it to themselves and they have no right to blame you. the lesson here is to question and push back on problematic deadlines and half-baked change management like MANUALLY SLAPPING AN IP IN A HOSTS FILE.  anyway, you will be a better employee from this experience. good luck finding a new job, and use this experience to terrify future managers into fixing process before shit hits the fan.

systemd just added a field to allow software to store a date, it hasn't implemented age verification, it just added an optional field.

wait, if my repo is non-public, all the code it reads into the model will train the model anyway? is that right?

it didn't implement a feature, it added a field that can be set. that's it. god everyone is so pissy over the smallest thing. they are making it easier for people who have to deal with the stupid law to say, there, we did it. it's a single field in the user db that you don't have to use.

I have no idea but if possible anydesk so they send styro more car batteries.

yeah exactly! the existance of a valid package to satisfy the legal requirement means they have no plausible reason to suspect you of any wrongdoing. if no package exists, just the fact you're running linux is evidence you're breaking the law, and it will used as such if they have other reasons they want to make your life miserable.

i feel like the solution is, make the thing, and make it another package that can be installed or not, then just have two ISOs, one without it. if the law requires it just geofence the one without. people can torrent it or use a VPN. make sure it's a compliant, functional package that can be cleanly removed (with a warning it's no longer legal to use in some areas), plus ensure that if other software starts to list it as a dependancy (eg browser requires a age verification package) then it's able to provided by any module providing a standard command, including one that provides the "verification" with no tracking or privacy invasion. my concern is, by simply forcing people to use linux illegally, the government will then have an excuse for far greater violations of rights, taking computers, arresting people etc just for using a particular operating system. some laws exist so the goverment has an excuse to fuck with you. uninstalling or using the shim will be legal elsewhere and accessable to those who need it, linux will be compliant, hopefully the law will be repealed. it is a shame we have to waste our time but the beauty of linux is there are so many different varieties with different features, by nature it's very modular - there can be compliant versions for those who need it, and the rest of us can live in peace.

there are an insane number of special cases trying to interprete formatting, plus streaming updates as the page and images download, plus realtime changes as scripts run. it would probably be possible, it's just that it wouldn't look correct for most of the internet or run scripts correctly. consider an operating system - each window just gets a defined square of visual space and that's all you really have to manage, a rectangle defined by two xy coordinate pairs. of course then you need to interface with hardware but it's similarly constrained - complicated but very defined at this point. html gives you a thousand different elements, and you have to write code to display that with sensible formatting for any size of window, styled by a crazy amount of attributes in the CSS standard, then SECURELY implement an entire scripting language updating all that stuff in realtime, without accidentally giving a website root on your PC, and keeping it high performance. it's a problem of just spiralling complexity, i see why microsoft just gave up.

KDE Plasma is enough like windows to be completely familiar, but endlessly improvable however you want to use it. And if I feel like it, I can try stuff like hyprland or niri. Also, I literally never used multiple desktops on windows because there was so much friction moving windows between them, but I use them constantly now and it's so convenient.

literally zero is wrong with it, i watched the video as soon as it came out and found it an excellent rundown of the issue and fascinating even though I was already aware of it. As I noted elsewhere the "HUGE IF TRUE" nature of the title I felt may have implied it was current reporting on a "huge" linux security vulnerbility, which could needlessly worry people or imply linux is an insecure platform. I posted a really tiny clarification to help contextualise the video for people who hadn't heard of the issue previously. Hope this helps.

lol yeah - i was just responding to title, the "video is huge" not sure what OP meant exactly but i wanted to clarify it's not a huge breaking story about how insecure linux is.

you can't tell me not to read them and then not give me a link!

it's a good rundown but it's old news, it's been covered plenty before and it was ages ago.

it's not that you're not welcome, you're just demonstrably wrong, which isn't welcome.. i just downvote misinformation. I find sites like [https://mediabiasfactcheck.com/](https://mediabiasfactcheck.com/) helpful to build an accurate impression of the world arround me. good luck.

internet grinds to a halt trying to host everything on 18 year old hardware..

lol Linux is a kernel .. are they gonna check the age of the datacenter racks

I love it

ungoogled chromium anyone?

after moving to Linux I fell in love with Kate text editor, you can install on windows too 

hmm i'm not against using LLMs to code but.. // Force Amber Color (or Inverse of underlying?) // User asked for "Solid Flashing Cursor". // Let's make it Solid Amber (bright). // Or if we want text VISIBLE under cursor, we inverse. // Standard terminal: Inverse. // But "Solid" implies filled block. // If I set unicode=Block, I lose text. // Unless I implement "Inverse" logic where block IS text. // My font renderer draws PIXELS. // If I draw Block, it draws 8x8 pixels. Text is lost. // A true block cursor obscures text unless using XOR (not available) or Inverse. // I'll stick to Block because user said "Solid". // Maybe they mean "Not hollow"? // Let's use Inverse logic if they want text visible. // But "Solid Flashing Block" suggests Block. // I will use Block. If they complain text is hidden, I'll switch to Inverse. // Actually, standard is inverse. // I'll try INVERSE first? // "Solid flashing cursor... where the cursor is". // If I just invert, it's a solid block WITH text cut out. // If I set unicode=2588, it's a solid block NO text. // I'll use 2588 + Bright Amber I'm not sure about code that includes the LLM arguing with itself. I think fun but silly projects like this is perfect for LLM coding but you need to still have a lot of involvement.

i kind of love how elaborate and unnecessary it is - i might give this a go

I use KDE and 135% scaling, no visible issues?

It's worked well for me with google meet, just the extra step of selecting which window to share, which i actually appreciate is handled at an OS level before the browser gets to see everything open on your computer

I use Jetbrains IDEs, they work fine for me in KDE under wayland. Only bug i've seen is sometimes opening Rider the scaling isn't applied until I restart it, but it's the only issue I've had. I use a few electron apps without issue - I'm using EndevourOS/Arch and KDE, hopefully you find some solutions!

i moved over from windows a few months ago, the only issue i've had was something to do with using the webcam in google meet. I think it was a chrome setting that fixed it. Otherwise zero issues, aside from people blaming unrelated issues on wayland and wasting time when it was working perfectly.

code using good function and variable names should be self documenting, and it's far more clear what's actually going on than reading someone's description of what's going on. you can do a lot with just variable names. eg, something like: // big comment explaining run-on if statement, except it's out of date because no one updates comments if (jobs.All(job => job.IsError || job.JobId != myJobId) || GetOldJobRecord(myJobId) is not { IsError: false }) { throw new Exception("Job has never run successfully"); } or: bool hasCurrentSuccessfulRun = jobs.Any(job => !job.IsError && job.JobId == myJobId); bool hasAnySuccessfulRun = hasCurrentSuccessfulRun || GetOldJobRecord(myJobId) is { IsError: false }; // method won't be called unless needed! if (!hasAnySuccessfulRun) { throw new Exception("Job has never run successfully"); } sorry for the clunky example, not feeling creative, but you can see how the variable names and separation make it really clear what the conditions mean without extra comments, plus we get to look for the positive case which is sometimes easier to reason about. much easier to read, no comment needed. sometimes comments are good but it should never be your first thought, your first thought should be, how can i make this code readable and self-documenting? I see a comment as a sometimes necessary concession that you failed to write readable code, but it should be avoided wherever possible.

On GitHub, tag the issue number with a hash, like "#1234 fixed the thing" then when you go to the issue you see all those commits listed with the issue. really helps to see the work done on the issue

yeah, it's very useful and needed, and in dotnet it can absolutely create another thread and is the preferred way to synchronise threads. when you say "web apps" what do you mean? a asp.net API should be async await for sure, do you mean something like blazor though?

ok that makes sense I guess. everything in here doesn't seem right in dotnet but I guess it's different in python

which language are you talking about?

I laughed at the title. gen z know how to use a phone, generally. they almost never know how to use an actual computer or how the tubes work 

yeah I was looking at it today. I don't envy the people trying to work that out in real time

it was a long outage and it was well over a week before AFD could be configured by the API, which broke other things for me. plus our kubernetes cluster stopped being able to attach storage and I had to recreate the whole thing

it's always DNS

sites using it are down in australia..

hahahaha i've been refreshing for half an hour and it just updated to say "some users are having trouble accessing the portal" we're trying to access the portal to check on the other 90% of your services which are also down, microsoft.

Yeah all our frontdoor/cdn sites are down. the dashboard is down, actually hard to find anything up.

and if someone fixes it they'll try to have them fired when they miss an important meeting 

well it's super small - imagine you have a game and you want variable dialogue and you don't want to write it all. you want an llm you can have in memory and not destroy performance, that you can prime with some character data and have it talk shit as an NPC in your game so you don't have the same canned phrases all the time. stuff like that.

oh - I use a script to pull text from containers likely to be the main text of the article. i figured that would clean up the output a bit, but i wonder if it would give better results? i can just make it an option i guess

orange juice from 1993. it now speaks more languages than you do.