Follow Slashdot blog updates by subscribing to our blog RSS feed


Forgot your password?
DEAL: For $25 - Add A Second Phone Number To Your Smartphone for life! Use promo code SLASHDOT25. Also, Slashdot's Facebook page has a chat bot now. Message it for stories and more. Check out the new SourceForge HTML5 Internet speed test! ×

Comment Violating comment prototocol to illustrate a point (Score 1) 474

Note how that last comment refers to "the average cloud-ready server"...

NO competent large scale developer would ever even think in terms of of a "cloud-ready server"! That's exactly what I meant by technological refactoring. It IS happening but we're not bothering to notice it. (Other than some fat-fingering maintenance at Amazon last week) We have uptime expectations, performance expectations, that were impossible a few years ago.

As younger generations of developers move in to replace older ones, the loss of implicit and limiting assumptions of the older ones will allow for newer ways of thinking about the problem space. That is where the stepwise improvements will come from, just at they have been arriving all along.

Comment Re: Market (Score 1) 474

I remember those days too... No stack, non-reentrant architecture, insufficient resources to emulate it in software (no base-indexed addressing like the IBM 360, for example). In fairness, the sorts of things we did with -8's were simple enough that the lack of resources was an acceptable tradeoff.

In response to the endless comments about bloated software, etc., expectations have increased either in step with, or perhaps ahead of, capacity improvements due to Moore's law. Being old enough to remember how much time was spent on a given "capability" in 1973, the godlike power granted to developers by a $10 pi-Zero-W, $35 Pi (quad 64 bit? way past Sci-Fi...) or Odroid is a wonder to behold. The average cloud-ready server is multiple orders of magnitude more powerful yet.

I would argue that we have no real sense of how dramatically today's world differs from those days, even those of us who were there. The word "inconceivable" just doesn't reach far enough into fantasy to compare with normal expectations. It is unsurprising that some deep technological refactoring is in order given the orders of magnitude differences between then and now.

There has been some paradigm-shifting between then and now, but more is needed to really take advantage of current technology. Until then improvements will look a bit anemic. I wonder whether I'll live long enough to see that shift.... It promises to be *very* interesting.

Comment Re:Business folk (Score 1) 723

From your response to Gravis Zero:

Business people in general have to compete with other businesses.

True for small businesses and individuals engaging in economic activity where the market tends to be far larger than the entities participating in it. A serious problem arises when a marketplace experiences consolidation in one side of the buyer/seller populations: I claim that conditions (where a single entity or a small number of entities gain control of one side of the market) attracts people who absolutely take improper advantage of people.

Several examples in our current economy:

  • Health insurance (rates increasing between 3 and 8 times core inflation)
  • College (faster than inflation, though I don't have specific numbers. In the '70's students could work part-time and in summer to cover the year's expenses; that is flatly impossible now.)
  • Lifesaving medical treatments (not the expensive new ones, I'm thinking of the ones off patent for years but still egregiously priced)
  • Laborer excess, such as in coal mining, where equipment and technology have decimated need for human labor while preserving margins for the owners

My wife and I have a rental house, which we think approximates a "dividend stock" requiring a lot of yard maintenance :-) but returning about 4-6% ROI depending on maintenance requirements in any given year.. The home rental market has not (yet) consolidated so badly that we get to compete with lots of other small-time owners in a market with lots of renters. It works well.

Then there's the energy market, the communication (ISP) market, the hospital system market in most cities, the health insurance market, manufacturing big ticket items, etc. Those large organizations are run by executives who know exactly how much market power they have, and they use it relentlessly.

Comment Re:Hardware better? Matter of judgment (Score 1) 115

...Great comments, especially WRT the UBEC. My work is always in a lab environment (or at home, where I use beefy USB chargers...), so it never occurred to me to look toward the RC world for an inexpensive DC-DC converter. Thank you for that!

The RPi is not designed for industrial application, and every one of the characteristics you cite, while being important for "pro-grade" product performance, also add cost that is unnecessary for 99.9% of Pi use-cases. Professional engineering time (that's us...) is precious, so we don't screw around with toys for delivered production hardware. The RPi, Odroids, Pines, Pi-clones, even the Beaglebone Black are all fabulous toys that we use to support diagnostics, lab automation, etc: If a $4.99 SD croaks we image a new one and we're back in business.

Industrial strength hardware (fanless industrial PC) costs about 20x the toy hardware: about 700-1000US but it's still fabulously cheap compared to the way it used to be. (I won't get into the ugly details of trying to talk my management into getting rid of VME hardware and 25 year old designs...)

Especially when combined with trick like RAM-resident root and tmp filesystems, EXT3/EXT4 on modern high-endurance flash is fine, as far as I can tell. I've run endurance checks of our current CF cards (1 or 2 gig, extended-endurance) and they are still just fine after 10-15 years of simulated activity. Our product uses flash only for logging and vehicle reporting. Modern wear-leveling and larger media let us forget the whole endurance question. NO more JFFS2! :-)

Comment Hardware better? Matter of judgment (Score 1) 115

Personal note: I have and use Odroid -U2, -U3, -C, -C1, -C2, RPI2, RPI3, and a UDOO (original backer), mostly as micro-servers. I don't require much customization and as long as that remains true, I find them to be great machines.

The Odroids are definitely better hardware, but the story gets more complicated when the question of kernels (and the binary blobs needed for media) are updated to mainline. I've heard but not verified that the original Exynos CPUs in the Odroid-Ux are supported by mainline kernels.

The Allwinner chips in the Pines, Banana Pis, Orange Pis, etc. lack complete HW docs and need critical binary blobs (At least the Allwinner H8 has long needed a DRAM controller library blob, for example). If Allwinner were to clean up their documentation and make truly complete hardware docs available, then the overall product would be better than RPi. Until then, RPi support is so much better than their competitors that it overwhelms the otherwise obvious performance advantages.

Quoting from

However, the support isn't complete for the Allwinner A64 and is blocked in part by lack of proper documentation. Andre commented, "Due to a lack of official documentation and hardware availability this doesn't go any further at this moment."

The Allwinner A64 is comprised of the less-powerful Cortex-A53 cores, supports H.264/H.265 video decoding, and is widely talked about as being the "$5 ARM SoC." Hopefully this mainline kernel support will get figured out in time for the Pine A64 shipping.

Comment Re:Undermining Faith in the Election (Score 1) 734

Read John Perkins' book "Confessions of an Economic Hitman".


His followup book "Hoodwinked" reports that the predatory nature of the bad actors from "Confessions" was so successful that many of those techniques are being used closer to home.

Look at some of the comments in this Slashdot article to see how well it works.

This society is being worked over, "big time".

Comment Re:JavaScript (Score 1) 295

RISC-V still hasn't yet designated different parts of their NOP space for trapping and non-trapping NOPs, so extension is going to be difficult, and the RISC-V Foundation still doesn't have a process for introducing extensions for review and standardisation. This is why the ARM ecosystem is so valuable and the MIPS ecosystem is a wasteland.

I remember ARM's earlier work, which had the very same difficulties you describe for MIPS architectures. Every implementer had a separate memory protection/paging architecture, every one had a different interrupt routing architecture. What a grinding mess that was... They wised up *just in time* (IMHO), establishing core ("Cortex") functionality which is required for all implementers. That stabilized the ARM ecosystem enough to allow it to grow dramatically.

If the implementers of RISC-V are wise, they would establish and agree on functional (powerful, effective) support structures such that it's possible for developers to implement kernels and OSes around it.

Given Softbank's purchase of ARM, I am watching their actions very closely. All of the embedded devices used in my company and many others are Cortex-M3, -M4F, and the many -A series processors. Softbank is in a position to increase their IP licensing fees for immediate profit, screw over the entire embedded marketplace, and would end up, after much consternation in the embedded world, ceding the market to competitive low-end X86 and one of the lesser-known architectures (MIPS: Microchip's choice for PIC32. Irony... :-) ). This is a good place to mention RISC-V *if* they are smart about portability.

Comment Re:Geeky magazine subscription (Score 1) 204

Mod up parent.

My father did me the favor of "leaving his Scientific American" mags around the house and a bored kid started looking at them.

At first it was very abstract and conceptual. It must have stuck, though. I will be 60 next month and still read every issue cover to cover and with special emphasis on quantum mechanics and cosmology.

Also: My uncle's gift of "Mathematics: A Human Endeavor", an introductory college text, probably "math for humanities students", but at 12 it opened the world.

Comment Re:Nope (Score 1) 468

I watched an old Twilight Zone episode on Netflix the other day (1966) which was about a factory where all the workers were being replaced by a computer and robotic system. They have been saying the same thing for 50 years. How many jobs today were not even imagined in 1966?

There are lots of new jobs, most of which are either beyond the capabilities of the people being displaced or which require several (5 or more) years of retraining during which time the displaced worker is starting over, earning nowhere near a sustainable income.

Comment Re: Extrapolation? (Score 1) 540

In today's libertarian paradise

I'll note such a thing doesn't exist.

You note correctly. In my US-centric view, which is where the big jobs question is being asked, people whose careers are being replaced are out of luck. And I am frustrated that here in the US especially, the people who are hit hardest by economic dislocation have the least resources to recover. The source of my frustration is that the benefits of technological development and world trade are shared broadly, while the costs are concentrated on those least able to find viable work in the new economy.

it's entirely possible to pay big bucks for entry into school

Education definitely is not a libertarian paradise. As to the rest of your US-centric criticism, where is the acknowledgement of the US government's decades old role as driver of inflating education costs at several times the rate of monetary inflation? That most definitely is not libertarian policy.

Libertarian philosophy is *explicitly* to "keep government out of the way", limiting government's role basically to defense and police work. Over the last 40 years, while demand for quality education has become strongly inelastic (it's necessary for individual economic survival, we'll pay through the nose to get it), the libertarians in our society have successfully reduced education funding at all levels, with college and grad school seeing the largest shift from public funding to students and families.

Libertarians in the US tend to be strict constructionists, and since education is not mentioned in the US Constitution, they believe it does not belong in the hands of federal government Cato Institute article. The states' management of education funding is inconsistent at best, and with the rise of "center-right" politics in the US, the states are reducing their funding too.

Thus my argument that the people who are dislocated by technological change rarely have the resources to restart at mid-life. Getting education is expensive, changing from skilled trade work to intellectual work is very difficult, and income while starting over is insufficient to cover expenses typical to a worker in mid-life.

It's late now, so I don't have the time to do a proper search for references... But I am pretty sure I've read that a high quality education policy tends to have a high return on public investment, particularly when it's properly managed.

Comment Re: Extrapolation? (Score 1) 540

The people who have dedicated large pieces of their lives learning to be effective in the jobs being shifted to automation have been, are, and will continue to be, fucked.

The TOTAL number of jobs may be only slightly reduced or even increased (though again usually somewhere else...), but the people whose careers are destroyed have no hope of recovery. So your argument is bullshit. Our society doesn't do a goddamned thing to mitigate the destruction that is focused on the people who have nowhere to turn.

I believe that those of us with privileged IQs and who learn for a living do not understand the flatly gut-wrenching transition from journeyman/master craftsman in a trade (machining, driving; choose a skill formerly limited to humans) to pre-junior apprentice intern. Interns *sometimes* make a minor stipend... The ex-tradesman is near or past middle age with children heading into college, a mortgage, car payment, medical costs, etc.

In today's libertarian paradise, it's entirely possible to pay big bucks for entry into school. There is no room for someone whose cost of living exceeds the pay of early stage career development. Those who don't have the wherewithal to pay up are "obviously slackers unwilling to invest properly in their own futures".

I have worked in high-tech places and in very low-tech places (trying to bring tech to improve business processes, etc.). Generally, a small fraction of the younger workers are prepared for the inevitable changes, but once past a certain age, that flexibility goes out the window as family and financial commitments place extremely tight restrictions on a worker's choices.

My family moved to an area near southeastern Ohio in 1972, and the community there was *already* depressed due to reduction in coal employment. That was before much of the current explosion in technology came along to make a bad situation into a complete disaster for the community.

Comment Re:Closures? (Score 1) 497

Sounds like a PDP-8.

I made the mistake of coding "JMS" instead of "JMP" in a disk boot routine exactly *once*, teaching an important lesson... Three days later, I had recovered most of the files on the RK-05.

The IBM 360 instruction set didn't have a hardware stack but the SAVE macro served well in its place.

Async (the select/poll model) and multithreading both have their place, the toolkit should match the application.

For the embedded world, I've discovered that run-to-completion "tasking" is so much easier these days, when the average microcontroller's clock speed is >= 120 MHz. The ARM Cortex world has such friendly exception handling that we simulate foreground/background "tasking" with ISRs.

Comment Re:2 more I've seen (Score 2) 497

Scaled floating point, unless you're working with an IBM Z series (does Power8 have decimal float?).

Every modern processor has fast floating point; a double can store and express 53 significant bits, so express the floats in pennies. Or if not in USA, the smallest denomination in the money system you use. Server class machines probably do long doubles.

("Every" includes all Raspberry Pi models, the Odroids, Banana PIs, Orange Pis, etc.; essentially every X86 since 1991, Every Power since about 2003 and all ARM since Cortex-A.) I haven't kept up with MIPS. Even the microcontroller world has single-precision floating point hardware in wide deployment: We use Kinetis parts (ST's parts have Cortex-M4F too), so easy 32-bit ints and floats is the order of the day. A reasonably structured interrupt service function (no assembly required, BTW) executes to completion in 1 microsec... It has truly changed the nature of our deeply embedded work.

The result: Use a commonly available toolkit competently.

Slashdot Top Deals

Riches cover a multitude of woes. -- Menander