Tesla is real and has been producing interesting products for years. Most of the products have not been cheap enough for consumer use. I'm figuring Musk plans to change that.

I never got on the Java train in the first place, python is where I live for initial versions, with followup sometimes in C if need be. Java is dead to me, but always was.

20 years go, models were based on surface weather stations, a few hundred points of data for a typical country. Satellite data was mostly pretty pictures for humans to gawk at. Now Satellite data assimilated into the analsysis that is used for computerized forecasts. Much higher resolution data also. Satellite is now the main source of improvement. weather RADAR data is currently also pretty pictures mostly. *nowcasting* is about assimilating RADAR which is similar or higher res than typical satellite data.

umm... if your enemy knows that, then your enemy could tell you to stop doing something in hopes that you will double down and waste ever more effort there. This is the problem with *never do that*... you should never give the enemy rules to predict you with ... uh... damn...

Fleshing out what people are saying about python for prototyping. Figuring out how to do the right thing, or how to do the thing right, is the hardest part of programming. The hardest part of that is understanding what code does, especially over the longer term, when it is handed off between people. Python code is usually 5x to 10x smaller than code in other languages for the same purpose, and a real focus is to be readable. It's optimizing the right thing, human cognitive load.

If, after you have figured out the right thing to do, you need it to go faster, then it can be quite straight-forward to substitute bits of it with other language implementations that use the logic explored by the python version. Often enough, if you figured out the right thing to do, the python version is fast enough.

mind you, python apps always leak... it's a real issue in production deployments.

yes. fork/exec works. Using multiple independent processes and trying to keep them as independent as possible will often result in higher performance than trying to keep a single application and applying mult-threading to it. We replaced a C application that used semaphores and IPC with a python one that just used the file system, and it ran between 10x and 100x faster. Things that were slowing the C version down: They wanted a single log to read, so they made a log process and every body had to post to that log using semaphones. Python version: every process has it's own log. C version, had a *router* process that would accept products from receivers, because folks didn't want multiple writers to an output queue. In the python version, all the input processes just write directly to the output *queue* (not really a queue anymore because order is no longer guaranteed, but wasn't important for application.)

Often multi-tasking/multi-threaded paradigms lead people into bad decisions when overall parallelism or application performance is the goal. Python's GiL puts people in the right mind set of starting from independent processes, and having people do their utmoste to minimized interaction points between processes, rather than resoort to multi-tasking/sychronization.

Multi-tasking is about synchronizing access to shared resources. Synchronization is another word for making processes wait. Waiting is bad if you can avoid it.

Windows subsystem for Linux is the thin edge of the wedge... gradually build more stuff to run on WSL, and then a few years from now, everything flips, and instead of running linux apps in a windows host, the base os will be linux, and the windows apps will be running in wine... and it will still cost >100$ to buy the *license* bute they can fire almost all their OS folks, and their cost will be nil. If they play their cards right, just us nerds will notice.

BS. Google thinks your information is way too valuable to sell it to advertisers. They don't share it. and that's a good thing. Google keeps all the information and use it to allow advertisers to target (using google's tools & API's, not the raw data.)

The dystopian stuff happens when the data is shared willy nilly, and there isn't any particular person in charge of the data: Facebook providing API's that allow open harvesting, Apps on IOS and Android that allow similar harvesting by random 1 or 2 person companies, and/or fronts for foreign entities. It is far more difficult to find hundreds or thousands of such organizations, whack-a-mole style and get them all to fix their behaviour. so to me, Facebook's business model is a problem. Apple and Google's app store policing is critical, but what Google does with data is pretty much the ideal. We are getting services in exchange for giving data to one large corporation that is amenable to pressure (the *one throat to choke* model of risk mitigation.)

iphone 5s, released in 2013, is still supported by IOS 12 in 2018. Even people who change phones every two years would prefer to be able to resell a functional device. A phone without updates is a brick to me.

I just happen to be working on something where I need a windows port of a python app, and enabling ssh on a windows box is great, and I then ssh into it from from ubuntu laptop, and I can use cmd, powershell, or even bash. great! Installed vim, but I can't use it, the curses library/screen refresh doesn't work properly. If I do man man, I get the first page, and then the following pages don't show up. refresh in vim doesn't work. if I start typing the whole window goes blank. totally unusable. any idea who/where I would ask?

Linus has always had great taste, but his priorities have rightly shifted. Linus is getting older. He is mortal. Not now, but he sees that at some point he will need to hand it off, and sees that that will go better if there is a healthy community to hand off to. So now setting an example for working well with others is now more important for him than it used to be.

Linux is undisputably, the most important kernel in the world. It is no longer in *startup* phase, and now weirdly part of the establishment. It isn't cool for the establishment to be cruel.

Firstly, a firewall is configured, as completely standard practice, to block incoming connections. Period. There is no additional exposure. IPv6 addresses are not *exposed to the internet* because there is no NAT. People still need to ask permission for things, because the default is to deny. NAT is not the same thing as a firewall. NAT is one way of implementing a firewall, but firewalls without NAT have existed forever and restrict traffic just as fully. An IPv6 corporate lan is not more exposed than an IPv4 one. The kinds of things you are talking about should be dealt with by governance, and dealing with professional security and networking staff. One runs services in particular zones, not under someone's desk. NAT has nothing to do with it. I've seen crappy network admins that will gladly poke NAT holes in the firewalls for whoever asked, and I will be the guy complaining about that, whether it is done via NAT or IPv6. We agree network anarchy is bad. but NAT doesn't prevent or even mitigate it.

Secondly, your argument seems to be that laptops don't need firewalls because they are on the corporate network. 10 minutes from now those same laptops will be at Starbucks, or Best Buy, or stolen and heading kekistan. Laptops need to be configured for a hostile network regardless. NAT doesn't change a thing about that.

yes, I know, I work in an enterprise where idiots use enterprise NAT and it is a pain in the ass, every single goddamn day. NAT has become a religious cult of security people that think it has magic protective powers. That attitude is not based on any reasonable reading of evidence, just become a sort of chant, and it causes major issues for enterprises all day every day, but the issues are invariably technical, so long to explain that management's eyes glaze over. It's sort of death of a thousand cuts, rather than any one thing. NAT is also one of the major reasons deploying in the cloud is often easier... because you're forced to use public addresses out there.

but here's the kicker: That's still totally irrelevant to the discussion of IPv6. You can do NAT over IPv6 just fine. There is no *you can't have NAT* on ipv6 clause. It will work, but it so dumb that sane people generally won't choose that. But if your whole argument is IPv6 is bad for privacy because no NAT, then that's not true. If people want, they can deploy NAT also. It's just mind bogglingly stupid to do that. It's like insisting on driving your car using reins and stirrups. Sure it can be made to work, but why?

Perhaps some natting is going for cell-phones, but in north america, at least, every home user gets a public IPv4 address that is quasi-static. The ISP's don't want to do NAT because it is too expensive for them (think of the poor routers doing the natting for a hundred thousand netflix streams.) Carrier-grade nat is very rare, to the point that I have never heard of it being used in North America. You're just wrong. It's bloody expensive to do right. and if the ISP's do it wrong they get downtime, which makes the peons restless.

I don't think you know what 'expose the ip addresses' means, or what a router does. When ISP's deploy IPv6, they do not apply NAT. The fact that an IPv6 address can access the internet does not mean it is *exposed* if it isn't NATTED. one-way traffic filtering, denying all inbound connections is absolutely standard, and no ipv6 addresses would be any more *exposed* than a natted IPv4. Using IPv6 doesn't mean not having a firewall, it means not translating addresses at the firewall, that's all. All the bog standard filtering still applies.