778 karma · joined January 23, 2011
Current: Staff Robotics Engineer at Chewy
Prior: Atlas @ Boston Dynamic; Robotics and Embedded Systems Engineering @ Amazon Robotics, Rethink Robotics, iRobot; Automation Engineering @ Ginkgo Bioworks.
Early Career: Electrical Engineer in Semiconductor Industry @ Applied Materials, Analog Devices, Allegro Microsystems
You call SetJmp, where you would want to put your try/catch blocks, then no matter how deep down some call stack you go, if an errir occurs, you can immediately jump back to where you called setjmp and handle it / report the error or whatever, rather than having to handle and bubble up the error by at every intermediate level, "returning" your way up the call stack.
Not trying to crap on the tech, but it's not so simple as we can merely go to the fab and get small quantities of full custom, 100% yielding chips, for 6k ea.
They are just saying that for the same power usage as previous tech, they get 15% more performance, or 30% power reduction while achieving same performance as previously acheivable."
In particular with precision Analog / mixed-signal / RF, there is a real recognition that there is no perfect Amplifier / Converter / Black Box X, that will satisfy all applications. That all abstractions (and inputs) are leaky, a wire is never just a wire, ground is never ground, noise exists, etc. That the reality is that there are myriad tradeoffs to be made depending on what you want to do, and the best way to be successful is to acknowledge that fact from the start without any bullshit; And as a supplier to tell you exactly what those real-world tradeoffs are and help guide you towards being successful.
To their credit the applications engineers and even sales folks when I was there were just as happy to sell you a $.05 "jellybean", as they were to sell the latest precision laser-trimmed FancyAmp if that's all you needed because they were like minded engineers too, and at the end of the day, building something that did the job it needed to was satisfying.
(Plus, as anyone whose done consumer/ high-volume electronics will tell you, BOM cost is it's own variable in the optimization space that can lead to some creative design choices!)
Yes they will try to point out interesting parts they make, but it's like the next level beyond the application notes and a great way to see some practical designs and applications of analog circuit theory in the real world.
https://www.analog.com/en/analog-dialogue
The "rarely asked questions" archive usually has some fun ideas in it, like a white noise generator based on johnson noise from a resistor (https://www.analog.com/en/analog-dialogue/raqs/raq-issue-154...), or a signal chain for doing analog measurements on a single GPIO pin on an FPGA, complete with verilog code (https://www.analog.com/en/analog-dialogue/raqs/raq-issue-169...)
Definitely happy to see some data sheet love ..chip companies are definitely in a whole different world when it comes to documentation, compared to the software world I'm now in!
Modular apartment buildings also exist and can be thrown together quickly and affordably.
Not to be rude, but the problem is likely going to take a bit more to solve than throwing shipping containers at it..
auto p = unique_ptr<Foo>(new Foo());
Totally works, and may be easier to remember if you dont like the "magic" parameter forwarding.
The parent article linked even mentioned that their inaccuracy was part of the reason in considering the ban.
There is a huge difference between banning "finding the person they are looking for", and banning the use of bad tech.
It's not that it isn't great when you do need it, but there are also plenty of cases where you don't need it and it's just unnecessary overhead.
Maybe:
- Don't deploy a system that has this many false alerts (and that is what even the TSA calls them, they are not interested in finding dense hair)
- Stop continuing to use this stuff and spend money on it if we aren't seeing any benefit, and are seeing demonstrable harm.
- Don't replace human judgement with blind procedure that must be followed, when that procedure depends on a known flawed technology that isn't providing any demonstrable benefit.
- Accept that even if you want this kind of tech to exist, it isn't there yet, and you aren't going to "fix it in code" by adding a race variable.
Bottom line is, nobody is forcing us to use these machines / procedures. We aren't stuck with them and have to figure out some way to patch the code or we're all doomed, so we don't have to be so reductive about how we fix it. We could just say, this was a failed idea and throw it the hell out.