Member #394180

Member Since: December 31, 2012

Country: United States

  • Nice post (as usual), but I’m really happy to see it done as text & graphics instead of a video. Much faster delivery of info, no audio needed, easier to use as a reference and easier to peruse at work. Thanks.

  • But I’d much rather program this in C++ than python. Is there some kind of option for that?

  • Nice talk, good explanation. Reminds me of the first time I heard it in high school. Back then, Ma Bell was still the one and only phone company and Bell Labs was the premier corporate engineering research facility in the nation, if not the world (I know, IBM would disagree). Anyway, BL made a series of science exploration kits that they would send to high school science teachers who would hand them out to students. Each kit contained a manual and parts, sort of like the SIK. The manual would explain the basic science, describe how to build the experimental gear and then suggested experiments to run with it.

    My kit was the solar cell kit. In addition to an explanation much like Pete’s, there was an electric heating coil, a silicon crystal wafer, several vials of chemicals, tools, wire and solder. You added your own refractory bricks (to make an electric furnace), safety goggles, gloves and soldering iron. Then you followed the instructions to make 3 solar cells from scratch (break the wafer, dope it by applying the chemicals and baking in the furnace, plate attachment points onto it and solder on the leads. Can you imagine the reaction today to handing this stuff to a 16 year-old? Toxic materials, 800-degree+ plus furnace?

    Thanks for the trip down memory lane.

  • I have one of these: Alt text

    It has 4 of these which let air in from the sides without letting vermin and rain in, too. Alt text

    It also has one of these on top to let the air flow out or get sucked in/out if you turn on the fan. It has a lid to keep out rain and a screen to keep out bugs. Alt text

    It’s powered by a 12-volt battery which is recharged by a power cord or a pair of these if you want to be cordless. I got 2 100-watt solar cells and a regulator for $200. Alt text

    If you’d like, I can do some tests to compare temperatures when everything is closed, when the vents are open and when the vents are open with the fans on.

  • Instead of fans and solar panels and controllers and stuff, I’d let Mother Nature do the work for me with a semi-passive convection system. Hot air rises, so a set of vents at the bottom of the car and vents at the top would allow the hot air to rise out the top vent and the cooler air to be drawn in from the presumably shaded area under the car. By using those bimetallic vents that open when it’s hot and close when it’s cool, no power or control needs to be applied.

    Of course, there’ll have to be some clever design to keep out water, car thieves and vermin, but the solar-powered fans need that, too.

    If one must use a solar panel, instead of a $90 3-watt panel, I’d use a $50 50-watt panel. That’s 17x as much power for about ½ price. That’ll get your fans really turning.

    I like the idea of no regulator, provided the fans can handle it. It much simplifies the system.

  • Which Python? It seems to me that the 2x/3x differences make them two different languages based on a common root (think American Southern English vs. British Cockney English).

    Anyway, it’s a Turing-complete language so it can eventually solve any computable problem. That makes it a “real” language, no matter what any haters may say.

    Duck typing is a crazy dangerous thing in any language, especially in hard real time apps. It also seems to me to violate “Explicit is better than implicit”. Since finding a defect gets approx 10x more expensive in each successive phase, I’d much rather have the language system catch the bug in the language translation phase than to have to wait until the execution phase (possibly while the app is in use for real) for the operator to notice a bad type.

    Soft machine-controlled garbage collection encourages sloppy data structure use with bad consequences for hard real time apps. This has been true from LISP to java to Python.

    Interpretive soft machines are slow (and occasionally buggy) in any language and especially unsuited for hard real time apps. They also have compatibility issues between different versions, sometimes unintentional, sometimes deliberate. I’ve seen this from UCSD Pascal to Python 2x/3x.

    I know there’s PyPy, but that’s a JIT. I want my code compiled far in advance so it can be checked, checksummed, packaged, etc. Cython does not get completely away from interpretation.

    The OO is nice to have, haven’t made up my mind on AO. The latter seems too much like throwing some of your code over the fence to be implemented in a less-robust language, but I haven’t used it enough to see if that’s actually true

    Overall, I’d have to say that Python’s really not for me, since I do mostly hard real time (in case you haven’t guessed), but if someone else likes it, I’m not going to complain.

    My comment wasn’t so much about Python, it was about adding another layer of interpretation to an already interpreted graphics package. I’d have said the same thing if it was a java or Visual BASIC or C# wrapper.

  • This is double the price I paid for 2 100-watt crystalline panels, mounting hardware and a 12-volt regulator. They work indoors and out.

    What exactly am I getting for the extra $200?

  • So why does the world need a python wrapper for Tcl/Tk? How many levels of interpretation can a poor GUI stand before it collapses under the weight?

    None of which is to say that there’s anything wrong with the article, just raising a philosophical point about the product. I believe it was in Knuth’s Art of Computer Programming that he mentioned a system running on 6-levels of interpretation (using 1970’s technology!).

  • That clean water AI project does a good job with the AI, but will fail miserably on the microscopy/microbiology side. For one thing, yeast is way bigger than pathogenic bacteria. The latter require magnifications in excess of 1000x for properly seeing their shapes, which in turn requires oil immersion lenses and specialized methods of preparing the slides.

    Then there’s the problem of concentration. A sample of water that can make you sick is not a teeming mass of bacteria. It’s lots of clean water with some bacteria in it. When you take a random sample, then sample that to make a slide, chances are that you’re going to see mostly clear water. That’s why typically you take the sample and grow it in a petri dish until there’s enough bacteria to test.

    Finally, a water treatment facility doesn’t actually have to identify the pathogenic bacteria, just make sure that they are all dead. This is done by unconditionally filtering and treating the water. This needs only enough analysis to show that the processes are working, not to actually identify the bacteria.

    It’s a similar situation with particulate matter. Enough filtering and settling will prevent particles from exiting the system.

    And for toxic solutions, chemical analysis is the only way to go. No microscope-based system will show clear liquid contaminants, no matter how good the AI behind it.

    So as good as the work that went into this project is, it really won’t do what it’s meant to and I’m puzzled that it placed at all, let alone got first place.

  • How about controlling the pattern based on the Qduino’s analog inputs? You could set up an acrylic light organ in no time.

No public wish lists :(