My bad, I thought they were moving from Apache to something more restrictive / less open (the way so many have recently, to effectively “source available”), especially by their wording — which conveys to me they’re frustrated they aren’t “capturing” the “value” of their code.

AGPL is not my favorite license but it has its purposes I suppose.

deleted by creator

Not too surprising, it takes a 100kW AI rack to accomplish a fraction of what I can wrt writing code, and I can run on tacos and diet coke.

SLC has a glut of qualified people that could staff these offices, the fact they only got a 4% COL raise this year tells you most of why they might have trouble keeping people. The COL of SLC has absolutely skyrocketed since 2019.

But that’s actually besides the point, you know the real joke about this, they say there are 15 open positions, yet when you search for dispatch job postings, they don’t list any, that’s from their own site – only if you dig through SLC’s specific job portal do you even find a single posting for dispatch.

Maybe they should spend less time on AI and more time trying to hire actual people.

But the point of CDNs is to direct connections to a geographically-near IP, yes?

That’s generally right enough, the goal of a CDN is to deliver content from a server close to the consumer as possible (ideally on their ISP network using cache servers to avoid going out over the “wider internet”.) – however CDN networks typically also use Anycast IP addresses, which means that all of the CDN servers across their network use the same pool of IP addresses, and BGP / the routing table dictate what actual physical server you get routed to. This is typically the ideal closest server, however sometimes you want certain IP pools in certain regions for legal (China), or technical reasons, so the IP address returned by a given A/AAAA lookup for a CDN isn’t a given. There’s also ECN and other optimization CDNs can do on the lookup side but that’s outside of the scope here.

The domain name that any CDN webserver in different regions will get in the HTTP request headers is going to be the same, CNAME or no.

Yeah, so the CNAME just says “whatever A/AAAA address that resolves to” and the HTTP client will send whatever HOST it thinks its connecting to, meaning you can’t “mask” the actual domain you’re using by using a CNAME record.

Technically if you have a totally static IP serving a single site, it’s possible to ignore the HOST field and always serve that site, since logically, any request is only meant for that given site (this is basically the default site on something like Apache).

My main point is that there’s really no getting around that CloudFlare requires you to be locked in to their platform even if you just wanna serve R2 files from a subdomain, and I personally find that a bit spooky, migrating nameservers can have very long propagation times leaving your site unreachable if they decide they don’t want you as a customer anymore, or as a shakedown.

The way CDNs and virtual hosts work in general is to read the host field in the HTTP header, otherwise unless you dedicate an IP for each domain / “web site” there would be no way to know what to serve.

The issue is if you put the CNAME of foo www.foo-cloudflare-cdn.com. then it will just resolve to whatever the A/AAAA record is for that, and send the host of www.foo.com – which they will only service if that domain is hosted with their nameservers (they run automated checks to make sure you’re actually doing so). So there isn’t really an easy way to just give cloudflare some subdomain, unless you pay them $$,$$$+ for the privilege.

Valve actually does that, ironically enough, for the steam community web assets they use Fastly, Akamai, and CloudFront, all on subdomains of course 🙃.

Cloudflare tries to enforce pretty strong vendor lock in by requiring you use their nameservers.

Also subdelegate domains are an “enterprise” feature, so no luck there.

Basically the CDN market sucks, not a shocker Netflix, Google, Valve, and many others operate their own.

It sucks because up until the “sales team” rugpull, they’re the cheapest (and closest to reality) for bandwidth cost, virtually all the other CDN providers charge astronomical prices and their margins are hundreds to thousands of percentage.

Google themselves don’t really follow material all that closely over their entire product line.

Android 6 was basically the peak of the UI, IMO, the icons were very consistent and nice early material.

In later versions they shrank the icons and stuffed them into circles and started using a horrible color scheme, then they killed blobmoji and started outright copying Apple’s hideous emojis with that awful gradient and pseudo-skeumorphic visuals.

I only “follow” because whatever Apple does gets broadcast by every media outlet in existence. Also Google started blindly following Apple design since they killed my beloved blob emojis.

Also not a fan of the critical UI elements being popped out into floating islands, very easy to accidentally hit underlying page content when there’s effectively zero padding around controls (on touch devices, as the ad companies have discovered by making the × icons smaller and smaller).

Frutiger Aero.

Usually is. Still common among network admins to hear dumb shit like IPv6 being less secure because no NAT. 🤦‍♂️

If it was a simple geoip lookup that isn’t really reliable wrt anycast addresses (or even addresses in general).

9.9.9.9 for example gets reported as Berkely, CA (US). Which is only partially accurate, for complicated business holding and ASN reasons, but is not representative of what DNS PoP you’re actually using at any given time.

Quad9 is a Swiss org, but it operates at hundreds of PoPs inside many different countries (anywhere PCH has a presence), their addresses are anycast so it’ll use whatever the upstream routes/BGP dictate.

Both Quad9 and CloudFlare have the closest DNS for my network, at around 1ms RTT. However CloudFlare doesn’t support ECS, so I use the alternate Quad9 service that does, since it gives me better performance on a number of CDNs.

Moderately, it’s still not as good as Play Music it replaced, and frankly the only reason I use it is because it comes with Premium (and Lite gets ads so fuck that deal), otherwise I’d subscribe to something else for music (aside from growing my album collection on Bandcamp).

Getting vertical video before modern codecs (AV1∨HEVC), and the same bitrate limitations since it was justin.tv.

It’s impressive how stagnant Twitch is, and how expensive it’s purported to be.

The ‘enthusiast’ side where all the university students and tinkerer devs reside is totally screwed up though. AMD is mirroring Nvidia’s VRAM cartel pricing when they have absolutely no reason to. It’s completely bonkers. AMD would be in a totally different place right now if they had sold 40GB/48GB 7900s for an extra $200 (instead of price matching an A6000).

Eh, the biggest issue here is that most (post-secondary) students probably just have a laptop for whatever small GPGPU learning they’re doing, which is overwhelmingly dominated by Nvidia. For grad students they’ll have access to the institution resources, which is also dominated by Nvidia (this has been a concerted effort).

Only a few that explicitly pursue AMD hardware will end up with it, but that also requires significant foundational work for the effort. So the easiest path for research is throw students at CUDA and Nvidia hardware.

Basically, Nvidia has entrenched itself in the research/educational space, and that space is slow moving (Java is still the de facto CS standard, with only slow movements to Python happening at some universities), so I don’t see much changing, unless AMD decides it’s very hungry and wants to chase the market.

Lower VRAM prices could help, but the truth is people and intuitions are willing to pay more (obviously) for plug and play.

That’s basically what I said in so many words. AMD is doing its own thing, if you want what Nvidia offers you’re gonna have to build it yourself. WRT pricing, I’m pretty sure AMD is typically a fraction of the price of Nvidia hardware on the enterprise side, from what I’ve read, but companies that have made that leap have been unhappy since AMD’s GPU enterprise offerings were so unreliable.

The biggest culprit from what I can gather is that AMD’s GPU firmware/software side is basically still ATI camped up in Markham, divorced from the rest of the company in Austin that is doing great work with their CPU-side.

Expounding, Nvidia has very deeply engrained itself in educational and research institutions. People learning GPU compute are being taught CUDA and Nvidia hardware. Researchers have access to farms of Nvidia chips.

AMD has basically gone the “build it and they will come” attitude, and the results to match.