StillBored

On linux do something like `systemd-run --user --scope -p MemoryHigh=XG -p MemoryMax=X+2G firefox` to constrain it since the firefox people refuse to implement the memory limits logic that say Edge has..

But at the same time, this isn't just a Firefox issue, although they may be worse than most. Its a fundamental tenant of GC languages. You trade CPU (for the garbage collection) for RAM utilization. IIRC This was apparent by the late 70s with Dijkstra, but very clearly indicated in say Hertz where they are explicitly scaling GC overhead vs memory footprint to find that it takes ~5x the ram to approach the efficiency of an explicit allocate/deallocate model. Lower footprints are achievable but the GC overhead goes way up just to reach 2x the ram footprint such that the GC tends to dominate the CPU utilization.

And there are lots of papers like this for long running processes. GC works well if your model is basically allocate as much as you like, exit the process before actually needing to do a proper GC pass. Especially for contended systems where the memory is returned to the OS. This is why say PHP tended not to have this problem, because the memory was largely freed at the end of every request.

Yes, but the contended case is often the critical part of an application’s performance. Weakly ordered CPUs therefore need fast barrier mechanisms, which in turn means re-creating much of the store-ordering, writeback buffering, and commit logic anyway.

Odin fails on a number of fronts. For example, integer overflow is simply defined behavior rather than some form of trapping/notification systems. This is no better than C really, its just defining what C tends to do anyway, and its a known foot gun. Pascal has footguns to, but they don't tend to be of the form "I added two strings/integers together and the result was not handled reasonably"

Recently dabbled in zig because I, like many people, want a "better C" that solves many of the footguns that C has (ex: null terminated strings, lack of integer overflow trapping, slightly higher level data structures) without all the complexity footguns of C++. But concluded its a miss too, I don't want all the dogma around packages, and symbol soup seemingly selected to be different than C. I largely want a "compiler" not a language ecosystem. Zig is seemingly forever going to suffer from chasing purpose built libraries. Aka, why bother with a zig formatter, or crypto when the first thing anyone is going to want to do is swap in openSSL/etc because its dynamically/runtime tuned for its target platform and contains that odd crypto code it turns out you need for pkcs11/whatever.

And the realization really is that Andrew Kelley's rant about go/rust/c, is answered by Pascal, which really is a "better C". Sure its verbose, but that I think is really a strong point in the end. Everyone has autocomplete editors and actually typing "begin end" instead of "{ }" is something your editor can be inserting automatically.

Also, calling windows API's was incredibly simple with delphi although almost unneeded, but the stdcall/cdecl decoration told the compiler to use C calling conventions instead of pascal.

I might be hallucinating, but IIRC this is now dependent on PCIe revision. More recent revisions have removed the slot power limits line, and so it works out that in theory a x4 slot (or any other size) can have a 75W load. IIRC this is stated in the middle of a giant section on up/down plugging.

(but its also displayed in the pcie PowerLimit capabilities you can see with lspci)

How accurate is it that the DoD has a handful of class A networks? And WTF do they need them for these days, they aren't actually advertising internal networks on the public internet are they?

Part of this is simply fixed by a tweak to copyright that if the IP isn't available for sale at a reasonable price the copyright no longer applies. Games, books, movies, music, etc.

I'm surprised that these articles never method Orthokeratology (Ortho-K), which are basically retainers for your eyes and reshape the cornea at night and wear off if not worn regularly.

I've considered it just based on that basis for the last 10 years but still haven't done it because I don't know a single person who has.

“640K ought to be enough for anybody.”

POE and existing wiring, and terminating copper on the lower end is dead simple for the kinds of people who wire houses, being able to run on cat3 phone cable would be even more of a bonus. There is a market for attaching APs, security cameras, and a load of other stuff on copper.

As an American who recently moved and can now get 1/2/5Gbit XGS-PON, in a location which is borderline rural/suburban and was originally platted out 50 years ago, at the same price I was paying for shitty 400/20... I don't think our failures to invest a single cent in infrastructure or regulation over the past few decades should define the Ethernet working group's priorities.

I think the larger point is that dumping baseband and going with OFDM/etc over wider spectrum allows those cat5e runs that are rolling off at 600Mhz (or whatever) and the super clean cat8/whatever to coexist with bad cables, bad termination, etc. The spec could easily be built for say 50Gbit, and fall back to 2.5Gbit/etc on 200M chicken wire runs.

Then the argument about "but we have to pull more cable to guarantee those speeds" or "It consumes to much power" all go away, and instead the analog side gets a bit more complex, but given the $100+ phy's in 10GbaseT the argument that it drives cost is bogus when triband Wifi7 USB nic's are $30.

Does the mac still lack a SMB/CIFS browser?

I was shocked years ago that the mac, famous for its early network peer discovery and zeroconf and all, couldn't present a list of SMB servers and shares despite that kind of function being around forever on every other platform in existence.

8051! I love it. Its like running web stacks on these ESP8266's without the crypto acceleration.

But at the same time, we have to stop pretending that 1Gbit Ethernet isn't utterly obsolete in the same way that RS-232 is. Useful maybe for low power, longish reach, but its slower than a good number of internet connections now, and the wifi on the other end too.

Ex: My house, turns out the 1Gbit uplink from the ISP provided hardware to my firewall was causing me to lose 300MB because it was actually provisioned at 1.3Gbit, and when I switched it to 5Gbit, my Wifi got faster.. Ex, I can get in excess of 1Gbit in about 2/3rds of my house now to sites on the internet.

1GbaseT is 27 year old technology this year, 10GbaseT is 20 this year, and by any other computing metric should be obsolete too since there has been a 25GbaseT spec for 10 years that no one has bothered to manufacture. And here in 2026, double or more should be easy with modern phy technology, and with proper line quality could easily be all of dynamic power, dynamic length and dynamic speeds over a range of cable types and length, both running at lower power and higher performance.

Its more than that, its that x86 vendors know how to maintain hardware backwards compatibility, they don't throw out the entire USB subsystem every time a new phy/whatever shows up because there is a standardized mailbox interface sitting in front of the actual HW. Same with the core platform, which works out of the box using 25+ year old firmware standards that are flexible enough to support simple sensors and behaviors, like lid close notification on a laptop for example across multiple OS's. Even something as simple as the firmware interface for handing off a frame buffer to the OS isn't universally support on arm platforms because a significant fraction don't support uefi. Apple was an early uefi adopter, but whatever internal politics they have, means they tossed even that on the latest mac's.