I've used smollm2:135m for projects in DBeaver building larger queries. The box it runs on is Intel HD 530 graphics with an old i5-6500T processor. Doesn't seem to really stress the CPU.
UPDATE: I apologize to the downvoter for not masochistically wanting to build a 1000 line bulk insert statement by hand.
That article is probably not the best way to support that idea though. It mentions "when 3.5% of its population actively mobilized against it" but doesn't explain what "actively mobilized" even means. It talks about how effective non-violence has been in other countries but then caveats that to being when an independent judiciary was present. It even uses Kilmar Abrego Garcia to support that idea, but fails to mention that a lower court's decision was ignored and the only reason the SC was involved was because the administration said it didn't have to listen to them.
Obstruction is good, but ultimately if you are not at risk of losing anything by that obstruction, it likely isn't an effective way to accomplish anything. That's even if you could consider it obstruction. If you are permitted to have a rally then you are not obstructing anything. You're just having a good time. Municipalities don't approve permits that obstruct, its the whole reason for permits.
Tailscale has the funnel command which exposes services like how you describe, but that's off the table.
Not quite sure I understand your layout, but if these are separate VPNs, you could run one from the server with a port forward (guessing that's not through Mullvad as they don't offer forwards any longer - to my knowledge) and then setup the general VPN on your router perhaps so you don't have to change ip routes for the whole network. You would still probably need to setup an ip route specific to the server VPN traffic on the router at that point, but that would probably be less work.
If this all being done from the same device then you would need to separate them out by IP routes.
I don't think its a matter of violence vs non-violence. Even in the samples provided by the article, its a matter of willingness to commit what would otherwise be criminal acts. Ghandi was successful not because of the Salt March but because they created the Declaration of Sovereignty and Self-rule and refused to pay taxes until negotiations were made.
I remember Penn and Teller did an episode that touched on this on a show they had. The big take away was there is a difference between doing good and doing something that makes you feel good. What's accomplished by a sit-in on a courthouse lawn on the weekend that you filed and received a permit to do from the city? People like to compare stuff like that to the 1960s civil rights movement, but here's the thing: Rosa Parks not giving up her seat wasn't a social faux pas, it was a criminal act in Alabama.
I don't think you'll get all of these points with the same tool.
You can login to your account and use recommendations with Revanced YT Music, but not download.
Freetube can't do #1 but can import your sub list and download.
yt-dlp can't do #1 but can download (mpv could be used as the player technically).
Invidious can't do #1 but can the others, but Youtube updated their API recently which largely breaks Invidious.
Terminal players like youtube-viewer, ytfzf, yewtube, etc can't do #1 but usually the others.
Mundane tasks weren't really the focus. This was a debate between Redhat and the Linux old guard where the points were all based on the extremes. They follow different ideas on how tools should work, though. Init systems focus on doing one or few things but doing them very well (the traditional UNIX approach). Systemd is a suite of many moving parts to accomplish a whole range of tasks (more modern). Init is mostly just bootstrap and services, but systemd is that plus networking, plus user sessions, plus logging, etc etc. More moving parts means increased complexity and more chance for failure. Systemd as a suite then becomes a potential single point failure where init based systems would not be. Scripting for either can be involved, but generally speaking init is/was easier to write things for.
I think most users today focus on Redhat's control and not putting too much faith in one setup for diversity's sake rather than the other points, but the original debate really was a philosophically based one. There isn't a right or wrong on these, but some really interesting history.
I think for those people it boils down to systemd being an init system that does more than an init system maybe should. Combine that with it being more complicated to work with and with Redhat not really being that open to feedback.
Honestly, most of your selling points while completely valid don't matter in this case I think. The problem is that is a repair business doing work for non-technical people and those are technical selling points. For example, my wife is allergic to tech. She wouldn't care about dual-booting or telemetry. She just wants the simplest possible solution that she doesn't have to think about. She's bored having to listen to me talk about projects/work and while she has to have a PC for daily life, that doesn't mean she wants to have to have it. She just needs it and needs it to be easy.
The biggest selling points to her would be:
It just works
She doesn't have to relearn things (meaning the layout and where to click on things)
It runs her stuff (literally all browser based applications)
Her files and pictures are there
That's it. I think the biggest positive sell to repair shop users would be "its just like Windows".
They don't need it to be better, they just need it to be the same.
Ok, this is your summarized argument: Accel is going to gut the company and run it into the ground because that's what they do, but they haven't ever done that, but they could, so they will, so that's the same as doing it, although they haven't, but it will happen in the end because that's what they do, but they don't.
Its not a strawman if what you say is in fact a weakly constructed idea. Its just a weakly constructed idea then. Its nothing but vague generalizations and "what ifs" you posted. Let me just put it this way: evidence or stfu.
So, even if Accel doesn't do that, which they haven't done that, they are still guilty of doing that. Ok, yeah. That's some solid irrefutable logic you got going there. I think I'll go back to arguing this with commenters who are a little less emotional and more grounded in real world points about the topic.
Historically, Accel has never pushed acquisition. On the contrary, they do the opposite. Its why they VC fund over 300 companies, but you've never heard of them. That's not to say they couldn't, but they haven't ever acted in that manner previously so logically it would be safe to assume that trend continues with Tailscale. I think that's important here: its not about ability its about intent. If as a organization you give funding to another organization (even non-profits) you exercise at least some control over them as they are dependent on that money to function. This is actually a point other commenters have made in regards to Headscale. Headscale is maintained by a Tailscale employee. As they fund him personally, they can exercise some control over him as he depends on that money/employment. Again, even their comments circle back to ability vs intent. Tailscale could influence their employee, but would they? That's where a lot of the VC argument goes. Its just speculation as what a group could do, not what they would do.
Tailscale builds on top of the Wireguard protocol, LF builds on top of (through grants/scholarships) the Linux OS. You can't argue that it doesn't matter that LF doesn't have control over the underlying technology, but then argue that it does matter in Tailscale's cause.
Firstly, I'm not trying to start a flame war with commenters, I genuinely just disagree on something and some people are getting a little hot under the collar by it. The Linux Foundation comment I made because ultimately VC touches more than people think. Even its something that isn't directly tied to VC, that money filters through groups like LF which is a non-profit and most would argue a quite legitimate organization. The point is there really is no separation or clear line of demarcation on what is "good" funding and what is "bad" funding.
Mordikan @ Mordikan @kbin.earth Posts 0Comments 92Joined 1 mo. ago
I've used smollm2:135m for projects in DBeaver building larger queries. The box it runs on is Intel HD 530 graphics with an old i5-6500T processor. Doesn't seem to really stress the CPU.
UPDATE: I apologize to the downvoter for not masochistically wanting to build a 1000 line bulk insert statement by hand.