
Pop the cork on the next 30
The traditional 30th anniversary gift is a pearl so here's wishing a pearl of a kernal.
I'll try it when the champagne runs out and the hangover cures kick in.
Linux overlord Linus Torvalds has released version 5.14 of the Linux kernel. "So I realize you must all still be busy with all the galas and fancy balls and all the other 30th anniversary events, but at some point you must be getting tired of the constant glitz, the fireworks, and the champagne," wrote Torvalds in his weekly …
Speaking of Intel, version 5.14 of the kernel adds more support for Chipzilla's Alder Lake platform that puts multiple core types onto a single die and prioritises workloads depending on their needs.
When version 5.14 of the kernel recognises and takes stealthy advantage of Intel platforms that put multiple core needs onto a single die prioritising workloads depending on the type of performance operation, are Spectre and Meltdown type attack vectors more enhanced and hardened/reinforced and re-engineered to react somewhat differently from the expected system programming norms.
I don't imagine Intel thinking that is additional support for their platforms though, whenever it renders their core source foundations compromisable, for it is a catastrophic systemic flaw in their structural base architecture which all but invites remote random renegade rogue access to cores and anonymous third party exploitation of executive functions.
Let's wait & see before casting stones, amfM. No point in going off half cocked. .... jake
Quite so, jake, there is no great rush. Time is something there is always a great deal more of to allow and encourage ...... Proper Preparation and Positive Planning Prevents Piss Poor Performance Permitting Prime Prize Plum Penetrations and Perfect Private Protocolled Pursuit of Public Parametered Projects and Pirate ProgramMING Productions for Pumping and Pimping in Presentations to Populations Puzzled by Progress and Prisonered with Pathetic Past Postings Profiling and Pioneering with Pilots Practised in Plush Promising Programs and Prolifically Painful Pogroms.
Psst . .... :-) some folk are not into casting stones, jake, whenever their forte and raison d'être is hurling GIGAntic FCUKing Rocks.
So the solution to Spectre is to run trusted code on trusted cores with trusted caches where everyone is good and totally trustworthy, where no one will be peeking at the caches. And then all the general riff-raff will be kept on the disreputable ghetto cores where they can only savage each other.
So more isolation using more cores is the optimal software workaround to what is really a hardware design flaw.
AMD/Intel/ARM should fix the problem in hardware like the LibreSoC people are doing:
https://libre-soc.org/3d_gpu/requirements_specification/
So in the last 30 years half a dozen (almost but not quite compatible) proprietary versions of Unix none having a market share greater than 20% have been replaced by half a dozen (almost but not quite compatible) Linux distros none of which have a market share greater than 20%. Well whoop-de-doo.
Just like 30 years ago they work great for servers (no argument there) where you assume from the get go you are now a sys admin. But despite the huge amount of work on the desktop versions they are just 90% solutions. They work OK until they dont. And now you are a sys admin wading through the desktop spaghetti on top of the server OS. NeWS anyone? So at best they are about as usable as Win3.0. In 1990.
So the only "successful" desktop Unix in the last 30 years has been MacOS X. Which kind of says it all.
But hey Linux is "free" so who cares about trivial stuff like that. As always, you get what you pay for.
And I suspect you, like me, have the technical skills of about < .0001% of desktop users. In my case its .00001%. I've been writing commercial application and system software for more than 35 years.
So OK for tech hobbiest and those with very high levels of technical skill. But thats it. And for the server guys its a great cheap OS where you dont have the licensing insanities of the old proprietary Unixes. But it takes a lot of effort to get a Linux configuration that gets a stability that was pretty standard on the proprietary boxes back in the 1980's.
There have been great strides in the desktop shells in the last ten years. But there is still that overwhelming sense just like all versions of Win before Win95 that the people responsible not only dont understanding GUI's but dont really like them. Hence all the Win 3.0 level stuff I'm still seeing
Every few years I give the current generation of distros a work out while trying to get real work done. At the moment building and testing a dev tool chain product under 64 bit Linux and various deployment targets. Sure the distro is a bit more usable than the last try out 5 years ago. But it soon fell apart as a usable system. The reason I write it off is not because I dont know where the problems are but because I know exactly where the problem are. Maybe in another 5 or 10 years it might be usable. But in the meantime I will be doing what most professional Linux devs do if the need a desktop *nix, use MacOS X to build.
To those of us who have written and shipped commercial software over the decades Linux is just another bloody stupid OS. Lots of them over the years. Its not some kind of cult religion, which it seems to be to so many people. OS's either help you get things done more easily or get in the way. Since I first fired up a Bourne Shell in 1981 *nix was always more of a getting in the away OS. When it came to writing shippable software.
But each to his own.
You sound like someone who didn't have the opportunity to work with the various flavors of Unix 30 or so years ago if you are trying to equate the differences between the flavors of Unix back then to the differences between the Linux distros. I had the fun of dealing with 6 different Unix flavors and I've used 9 different Linux distros. No comparison.