yet another gitea/forgejo instance with some extra plumbing, hosted in the EU. this is a github alternative, not a git alternative. rickub's conflation of the two seems deliberate.
100% agree, and no OS should be forced to implement this "child-locked" signal. the existence of OSs that do implement it should be enough (if you want to lock your child's device, use a lockable OS).
fun fact: the first mass-produced consumer in-car live navigation (1981) was all dead reckoning and gyroscopes. at this point, the GPS network was operational but not available for consumer use until 1988.
what? my impression is that it's truer than ever and this is why so many open-source projects are limiting ai contributions. and the reason GitHub can't seem to stay online for more than a day.
oh hello, thank you for your work on an excellent product! one small difference i ran into last weekend is that foregjo forwards the FORGEJO_TOKEN in the environment to composite actions so that they don't need to have a required "token" input.
no; gitea and foregjo both support actions-style CI/CD and both serve mostly GH-compatible APIs and have GH-style apps. it's not 100% compatible (forgejo is a little more compatible than gitea [^1]), but many of your workflows might "just work" without even renaming the .github folder.
^1: in my recent experience, for my particular use cases
my least favorite part of flying is the part where i'm strapped to a chair during taxi with a screen and speaker inches from my face and they use that moment to force a few ads into my eye socket. it feels very Citizen Kane.
pumping gas is another one that takes advantage of captivity to force ads. i hate it.
the argument is that they should be called traffic surveillance cameras.
for the record: whatever they're called, and in fact even if they just POSTed txt license plate numbers, i'd still be opposed because they're collecting info about people who've done nothing wrong (unlike red light cameras).
i see it more like an argument against increasingly networked solutions. red-light cameras used to just upload when an incident occurred (or maybe required routine physical downloading?). the object itself did one thing. now it's cheap to feed continuous live feeds to a centralized processor... and once it's there, it's cheap to do a lot more.
letting companies train on books for only the price of a paperback and then forcing them to destroy their scans would truly be the worst of all worlds. (ps: the scanning process typically destroys the book too; they chop off the spines.)
also, as this scales, what would this mean for closed-weight hosted models? i imagine it's possible (but difficult) to re-derive model weights by de-lidding and inspecting the die... so will this only ever be used for open-weights models?
not OP but yes, i consider the tech sphere's dive into mass surveillance to be war profiteering and have eliminated or minimized several in my life including amazon and spotify.
is this also true when the irrelevant details are (by coincidence) interesting (to you)? it sounds to me like GP is interested in mario kart and you simply aren't, so this is a perfectly natural result.
for real. if you want to make the web safer for children, start by holding ad networks to a higher accountability. it's painfully obvious "protect the children" is not the real goal.
wrapping text at 55-75 characters per line is often recommended in digital typography by UX. research implies that an optimum length exists to maximize reading speed, and although it depends on many factors 55-75 is often parroted as a universal rule.
as i understand it: that's in-band ECC and if you CPU/mobo support it, you can enable IBECC even on non-ECC RAM. IBECC has overhead, though. ECC RAM has on-die error correction, with corrections taking place before the motherboard ever sees it (and without the performance overhead of IBECC).
the thing about ECC memory is that it doesn't support reporting of ECC errors, so it ends up just hiding the fact that your RAM has started to go bad. this is good if you replace RAM every n years, but not if you intend to use the RAM until it starts degrading.
you're right, i misspoke by saying it's a GPU since it doesn't have video output. the just-in-time data pipelining wouldn't work for graphics workloads anyway.
i'm also interested in tenstorrent. they're building GPUs with cheap GDDR6 using a fast SRAM cache, and writing their own compiler stack (used instead of CUDA) that pipelines data to the SRAM ahead-of-time so you (in theory) never need to suffer the slow speed of GDDR6 for AI workloads. also they've got built-in SFP cages where the video ports would normally be.
yeah, i've been using cheap mechanical analog watches and wouldn't trust it to be accurate to-the-minute by the end of the day anyway. i kind of enjoy knowing only the approximate time.
another happy user here too. having openwrt with all features working and no tinkering out of the box (since it's their reference target) is a dream. this, plus the warm fuzzies of buying open-source, makes it worth it to me even with the 1GBps limitation and outdated WiFi (i use a separate AP anyway, like you). i swapped my ports in software to have 1GB WAN and 2.5GB LAN (which also lets me power the router with PoE, which i have coming in on the LAN port).
i wish there was a system like strix halo, but with enough lanes for a dedicated PCIe 5.0 x16 slot so you can have the best of both worlds: large sparse models on CPU with unified memory, dense models on GPU with real tensors and higher bandwidth memory.
i agree, with a bit more nuance. it's one thing for a government agency staffed by educational professionals (like a public school district) to issue guidance (or even rules) on how libraries should be stocked -- but what's become widespread is legislative bodies bypassing those existing structures and issuing laws without input from educational professionals.