And surgeons responsible for cutting open live human beings should be capable of not leaving tools in the person they're operating on, but it still happens. Professionals make mistakes. Garbage collection is a useful tool to make it more difficult to screw up.
If only there were some kind of portable machine one could use in order to look for metal left in a patients body...
If only there were a requirement to do things like count gauze pads before and after surgeries, and then account for any numerical discrepancy as being ON (not IN) the patient, and/or going into the biohazard disposal unit. You know: a documented procedure.
If you can't handle explicit memory management, then you can't write in MISRA-C, and if you can't write in MISRA-C, you should probably be kept as far as possible away from coding on life support systems, because you are not going to be good at it.
The idea that the could "run out of memory" in the first place implies heap allocation, at the very least, which is prohibited in the MISRA-C standard.
You may remember the Toyota acceleration bug in the ECC; it was due to a practice which, had they been enforcing their own coding standards (which were a partial intersection with the MISRA-C coding standard, which was under development and nearing completion at the time), they would have caught the bug before it started ending up in dead people.
What is heretical about novel new genes arising out of junk DNA? Molecular biologists have known for many years that so-called "junk" DNA played a number of roles; regulatory, and that most certainly novel genes could arise.
There is a small percentage of biologists which really, really would like acquired traits to be heritable (as a work study job in college, I worked in the lab of one of them, and we were cautioned not to talk about his theory outside the lab).
Every so often, one of the proponents of the idea of heritable changes due to environmental pressure, or more formally, either Lysenkoism or Lamarckism, tries to find a mechanism that could make it work. Even though it's never been demonstrated (the biologist in the lab I worked at was attempting experiments with, among other things, chelodina longicolis diets, to force physical changes, which he hoped the offspring would inherit, even though not on that diet).
This theory is what's known as "soft inheritance".
The main premise for its development in the first place was that Joseph Stalin was all pissy about genetic being a non-Soviet idea, and wanted an nice, Soviet alternative that better fit the ideology he put forth. This actually influenced a lot of decisions in Soviet agriculture that didn't work out badly enough that they ended up importing wheat from the West.
The last go-round was trying to use introns as a mechanism whereby he introns were involved in making traits heritable (and before that, it was endogenous viruses, such as PERV - Porcine Endogenous Retro Virus). Those were the main ones. The've also tried to explain it with varying degrees of gene methylation, and so on. Todays flavor is non-coding DNA (the correct scientific name for "junk genes").
Unless the can demo it in plants, mice, or fruit flies, etc., don't expect that the idea will go anywhere directly.
The sad part is, if they had concentrated on the punctuated equalibria model, which the article mentions, instead of trying to explain it as a short scale inheritable phenomenon, the might have had a really great argument.
(Yes, I am in the 90% who are skeptical about this, without further evidence and perhaps a demo).
Sorry to tell you, but they are 13 years late to the party.
A Foresight.org project on 2002 was the release of Ka-Ping Yee's CritSuite.
It was also "invented" independently as "Third Voice" in 2004.
It's cute that they believe they've done something unique with the grant dollars (that they should probably be actually spending on dealing with climate change, rather than beating reporters over the head with foam "pool noodles"), but what they've done is far from new.
Funny how Slashdot's fierce Fighters for Privacy turn a blind eye when it's something they don't care about that gets hacked.
I think we are just laughing our asses off about trying to decide whether Pat Robertson was actually able to employ competent system crackers, or whether the security on the site was so lame, given its purported purpose and clientele.
Also, note that this isn't a privacy issue, it's a database hack; it's just incidental that it involves a website that contained prurient details of illicit affairs, rather than, for example, credit card data for everyone who used a credit card at Target, Home Depot, and Lucky's, or the OPM's personnel database and security clearance validation records (I still think the goal was data insertion on that hack, rather than obtaining the data, or it was at least a dual goal).
If you think that information about sexual exploits, however obtained, should not be dumped en masse to the press, then you must be from Europe and it's "right to have your hookers dressed as Nazi photos be forgotten" laws. Because in the U.S. -- once it's out, it's out.
I'm waiting for the data mine on specific famous persons and political figures to be published with quite some interest... wonder if anyone is going to be dropping out of the presidential race over it...
"Jared Newman investigates why Flash won't disappear from the web anytime soon."
Because the media companies won't spend money on the alternative..
That's actually not true.
They are willing to spend a crap-ton of money on the alternative. It's just neither of the two competing HTML standards camps is willing to implement their draconian DRM demands for them, or deal with their "commercial insertion problem", other than making it a server side problem, where it has to be placed into the content stream, and the user of the browser then has the option of skipping it by jumping ahead in the buffer.
No forced commercials means they will not be moving their content to HTML5 soon.
NB: One company was even demanding that they be able to implement "pause on mute" within the streaming protocol, so that not only did you have to *watch* the commercial to watch the next section of content, you had to *hear* it as well.
Most of the sites offering television content -- Hulu, NBC, etc. -- use a video player and content library management system from a company in Orem, Utah, in order to implement their commercial insertion mechanisms in order to monetize their online content. This player only works in flash, and simply would not work in HTML5, even if you wanted it to, without giving people the ability to block commercials from these sites.
Other companies content is managed, and the DRM protection mechanism, and operational ticketing mechanism, is implemented using Adobe Access. Adobe Access is a content management system based on an older product known as Adobe Flash Access (actually, it's the same product, renamed). The content control mechanism therefore requires the use of a flash player in order to implement the DRM. Companies using this content platform include Google (Google Play) and Amazon.
From a non-content perspective, a number of companies do a couple of things which are impossible to do in JavaScript, but which are often highly convenient as UI features, and flash allows a click bypass in the UI for these companies in order to implement their desired behavior. One of these is GMail, which has an upload drop-zone mechanism for email attachments (actually, lots of web mail applications utilize flash specifically for this feature). Another is the Amazon "one click purchase", which would otherwise be "two click purchase".
Finally, a lot of content is simply authored with Adobe Content Creator tools. Say what you will about flash, but Adobe has the finest web content authoring tools, content management system, and so on in the business. Others are at best a poor second, and none of them contains all of the capabilities of Adobe Creative. Sadly, that lets them, as part of their content creation, transparently insert flash into your web content. You can disable this (there are some obscure settings to allow it), but doing so disables some of the features you can use on your web sites. There are more modern methods of adding those features back in, but Adobe has not provided alternative implementations, since they feel "the flash implementation is good enough".
That explains everything that I'd be personally "reluctant to discuss" about why my site was continuing to require flash. As you can see, there are reasons, and they aren't great, but it's not like everyone is using flash to crap "super cookies"; there are other ways to track you.
Yet another article that assumes Beijing = China. Sigh. It's like there's only one city in China. Imagine if European journalists assumed New York City was all there was to know about the entire USA. And China is even bigger, and has four times the population! I think the problem is due to the fact that most Western journalists live in Beijing, and they are not really interested in reporting about anywhere else other than where they live.
It's pretty bad in most areas of China where there are actually monitoring stations (which is where there are actually people). Here's a pointer to an interactive map which demonstrates it graphically
One has to wonder what the hell is going on in Kashi and diqu zhan Hotan, which are near the Kyrgystan border, and have the highest and second highest (respectively) "bad" numbers of any reporting stations in the world.
This is called closed-mindedness and provincialism if it occurs in rural people, but now it's suddenly acceptable?
Actually, it's called "journalists are assigned "minders" and are only permitted to go wherever the heck the government lets them go, and nowhere else, so they only see what the government allows them to see". Welcome to China; new employee orientation for the state controlled media for foreign journalists is on alternate Tuesdays.
When I lived in the Bay Area, there was a fire at a recycling plant that caused some reading to go 400, or 500 or something in Redwood City. It might have been particulates, not sure.
Yeah, that's a teensy bit down from Hiller. There was some suspicion of them not being able to handle the recycling load, and "accidentally" setting the materials on fire (the plant itself was untouched). Other theories included spontaneous combustion due to thermal rise during decomposition (only it was mostly paper).
It was particulates in the 76 or so today, due to smoke from the wild fires (which are actually pretty far away). Everyone got an emergency services robocall. Most places in China are about that, according to the monitoring mapping service (aqicn.org), but there are some that are running about double. Highest I saw was a 953 on the China/Kyrgystan border (kinda insane), and a couple real hotspots around Beijing.
I found it interesting that they shut down the San Francisco station (it must have been showing numbers that were unfavorable to San Francisco tourism). Worst in the U.S. is Medford Oregon; most other hot spots are in Washington State. There's a 229 in the Czech Republic. Russia has exactly one monitoring station; I'm going to guess it reports whatever Putin wants.
If you are interested in the world map (navigable Google Maps map), it's here (I left it centered on China): http://aqicn.org/map/californi...
Thanks, insightful comment. I'd like to add little more about how Windows handles USB devices though, as it is somewhat relevant here.
Thanks; interesting tidbit about the Windows model; it makes sense, but as I said, I rarely see devices in the wild with actual unique serial numbers. Port topology is, as I said, also used by Apple, and you can be in trouble if you change it while the machine is asleep and later wake it up. Mac OS X has other issues with sleep/move/wake that can be pretty nasty; kind of off topic (not USB related at all), but they will all self-resolve in at most 3-5 minutes.
I've noticed that the FTDI driver can break USB serial number reporting under certain circumstances too. I'm still narrowing it down, but USB 3 ports seem to be worst affected for some reason.
The Windows way seems like a reasonable way to handle a bad situation.
FTDI devices are really problematic. On USB 3, they tend to trigger "downstream charging port mode", and USB 3 frequently has a hard time recovering from that (which is why when you plug a Samsung phone into a Samsung computer, it will often charge, but then you can't talk to it with the computer, until it's fully charged, or you replug it frequently. FTDI also has driver problems in that the driver can try to run if you unplug the device, and if you did a teardown, panic either Mac OS X or Linux machines. Both OS's have written their own non-vendor drivers to deal with the issue.
Typically, I just put the FTDI serial dongles at the very end of my port hierarchy/topology, and then they don't do the "problem child" thing and screw with your other devices.
So well you posted a lot of text, and many people think you are quite insightful, you do not actually tell us what the problem is (apart from referring to the independent keyboards issue which is/a completely different/ issue).
Linux has had, via udev and systemd persistent device nodes for years [...]
They are not persistent for pluggable devices for which you have not made specific configuration file changes to the udev.conf for the device.
In addition to the explicit conf changes, you may also have to make a udev.rules entry, such as:
as a pre-run rule to remove the old device so you don't end up with a new device node with a different name/minor replacing it. You have to explicitly know the name "whatever" of the device in question.
In Chrome OS we had to add a rule like this for the Synaptics Touchpad on the Alex (Samsung) Chromebooks so that the device would not disappear on upgrades, due to the timing of rebuilding the kernel cache.
So yes, I did tell you what's wrong; no, you are not correct about it being ACPI -- that's an unrelated Linux religious issue; no, this is not my first rodeo: I do know what I'm talking about here.
because he is actually wrong. He has some personal problem with the USB implementation I think. But I think he is either wrong, or fails to describe the problem properly. In any case, its just a minor issue, ACPI issues are a far greater source of pain with resume.
ACPI is not the problem with the controllers going out, unless you are claiming the reason the USB devices are not coming back is due to a failure of the ACPI code to repower the EHCI/UHCI components?
Yes, there are major suspend/resume problems around the use of ACPI power states to implement suspend/resume. I really don't care about those, since those are "do as I do, not as I say" problems; they are due to expecting the ACPI implementation to religiously adhere to the specification so that you can push the envelope, vs. "do what Microsoft Windows does, and don't push it".
Pragmatically speaking, expecting religious adherence to the ACPI specification, instead of just doing what Windows does, is a moral high ground that will get you technically correct code and a non-working system. FreeBSD does a lot better about getting this right, but is still not as good at it as Windows, since vendors only care about working with Windows.
This is similar to Linux being all religious about UEFI and ExitBootServices(); there's a static library portion which doesn't have a UEFI mapping in the Intel reference implementation (which *everyone* uses), and so you if you call that, and then don't leave the recovered memory alone anyway, there are certain EFI calls that, if you make them, even though they are supposed to be available runtime, you are screwed.
When we are looking at USB devices is it possible this could be handled later in the boot process by having all devices re-polled and connected as new devices? So on resume previously connected devices are simply treated as new devices. It would add a delay to the resume process and I'm guessing there may be device confusion if you have 2 controllers attached, ie player 1s controller becomes players 2s after a suspend resume. But would this not be a simple, but manky, solution to the problem?
If you are talking about the application programs closing and reopening the devices, then yes, the controllers could be handled that way. You would have to do it in every single application, however, for every single device which was capable of moving around that way, so it would include audio devices, cameras, game controllers, and so on.
Further, you'd have to have a mechanism for notification of the applications that they need to close and reopen (and reinitialize) their devices.
This really can't be done at a system level, because on resume the devices look like arriving new devices, and the nodes are in use by existing opens. So it is, for example, easy to "lose" a built-in camera or keyboard or audio device, because those end points are in use, and the "new" device gets assigned a new end point.
One of the things I fixed in the Chrome OS testing environment was that the cameras got "lost" during factory testing on a number of devices where they were interfaces via USB (but were built into the clam shell). This meant that the test harness, which set up devices up front, would lose the devices because there were "there and open" in the Python code, but "not used yet before the suspend/resume cycle".
The fix was to move the camera open/close code, ant to modify the code to search for specific attributes which meant "camera attached to this node", rather than "node for a camera".
Technically, suspend is not the problem. The problem is resume. It doesn't probe and reattach the controllers to the same point in the device tree that they were in when the system was suspended. Since those are the device nodes that SteamOS has open at the time of the suspend, and they route to The Noplace(tm), the controllers become unresponsive.
This is a general problem in the Linux device model, and you can see problem in the device model poke their heads up in various places.
For example, if you plug in two USB keyboards, what you are going to see when you hit the caps lock key on one of them is that the caps get locked for both of them, but the LED indicating that the caps lock is on is only lit on the original keyboard. This is great fun, when you see the LED state on both keyboards relative to the caps lock state, when you bounce the caps lock key on the second keyboard.
Presumably, this was implemented this way in order to allow multihead operation, with the keyboards separately (and explicitly -- which they are not by default -- to separately running console instances. But it's indicative of deeper problems in the model.
The reattach operation should (in theory) be handled by the udev mechanism, which I'm told is subsumed in systemd; however, they've faithfully reproduced the problems in the original implementation in the replacement as well (so it actually doesn't matter which one SteamOS is using, so please don't argue about that crap).
There are two implementations that do work; however, they in fact work differently.
The first implementation is in Windows; it works by directly assigning the device descriptor based on the USB vendor and device ID. What this effectively means is that when you resume, everything gets the same descriptor (or a replacement) that it had going own. The keyboard "problem" is handled in Windows by making them "the same device" -- in other words, you caps lock on one keyboard, it does it on both, and you undo it, and it does it on both.
The second implementation is Mac OS X; it's handled by explicit enumeration order of USB bus devices, and the using the USB vendor and device ID *and* the enumeration position ("bus ID) to uniquely identify the device.
You'll also notice, if you look, that Linux has problems with keyboard internationalization and locale. This is most easily seen by using a locale specific keyboard, and having it not be recognized. Further, you'll notice that the character set differences are handled by tables in X and Weyland, and these are not the same as the console tables (i.e. the USB ID of the keys is not propagated through the full input stack, and there's a difference in operation between sending the events up through the console, vs. sending events up through the X Server). To fix this would require moving the HID key value translations into the kernel keyboard driver, rather than having it (mostly) in user space in all three instances (X, Weyland, console).
Finally, Apple is one of the few vendors that actually correctly fills out the USB device serial number field correctly, so it's hard to use that as a unique identifier (specifically, it makes it really hard to mask program all your controller chips, without adding a "burn the fusable links" step).
Further, they are also one of the few vendors that sets the locale field in their USB devices (most laptop vendors will get this right during manufacturing, by placing the value into the BIOS, since the laptop keyboard is actually matrix decoded by the EC via a grid hooked up to GPIO pins, and then the EC pretends it's an 8051 with a PS/2 interface for the keyboard to mux a PS/2 trackpad -- e.g. like a Synaptics -- to look like a standard PS/2 keyboard and mouse).
Apple handles for unrecognized keyboards by having you press "the key right of the left shift key" and "the key to the left of the right shift key", and then uses the key IDs. Not ure if they have a patent on it, but it's a lot more clever than what Linux or the BSDs do to
Reach your data cap up to 40X faster!
on
The Promise of 5G
·
· Score: 2
How much reading and writing is necessary for an average person to succeed. Donal Trump speaks at an elementary school level.
This is actually a valuable political and business skill. It's called "being able to speak to your audience", and it's fundamental to being able to communicate your point to people at that skill level themselves. Reagan had this same skill, as did Kennedy. You will find that all the people we consider "great orators" had this skill. Jimmy Carter, to a lesser degree (or he would have had a second term); surely people remember his "I want to be your president, because I like you". Mr. Rogers level dialogue.
I personally found both Ross Perot, and later, Steve Forbes, both intelligent and insightful about the problems we face as a nation, and they laid out their plans in great detail. But they both talked well over most of their audience's heads, which rendered them complete unelectable.
I don't know anyone in the Democratic field at this point who has a similar skill set to Trump. Perhaps Bernie Sanders. Definitely not Hilary Clinton.
Do not underestimate the value of speaking to your audience.
First, there is no such thing as a "rock star front-end developer". The front-end space is actually constrained enough that it's possible to know all of it, and act within the limits of those constraints all the way up to the boundaries and no further. So it's a pretty rote work position for a developer.
Second, their definition of what constitutes a "rock star" is inane, in that it's a stereotype of the behaviours of people who have achieved "rock stardom", and not a description of their capabilities. The issue is one of capability, and a "rock star" musician, like it or not, is capable of doing things which others are not capable of doing. It's rather as simple as that, and hence the migration of the analogy into other areas of human endeavor.
Having a developer who is a "rock star" is a significant competitive advantage, in that they will be capable of doing things which others are not capable of doing. This is a competitive advantage, in that it acts as a barrier to entry to your market, because it means that your competitor simply can not hire someone that is capable of competing on your level no matter how much they pay, short of hiring your "rock star" away from you.
This is the equivalent of "first mover advantage", without the pressure of having to execute quickly in order to maintain that advantage. You don't have to spend crap-tons of money on "ramp speed" and "burn rate" and "time to market", and "runway"; as a result, the VC feeding frenzy ends up owning less of a chunk of your company, and you get to be rich instead of making *them* rich (or while *also* making them rich, but not as rich).
So yes, it pays to hire "rock stars" for strategic things (and again, front end web design is strategic, but from end web coding is not; designers can be "rock stars", but front-end developers can not).
So yeah: there are some stupid categories in which to advertise that the person you hire be a "rock star"; there are many others where it's not actually silly, it's smart, or it's even imperative to have one.
Yes: it's really frustrating to recruiters, and to job sites like DICE, when someone asks for a "rock star", and they are incapable of delivering one to their customer, because they have none in inventory. Lump it. Be more desirable to "rock stars", and you will be able to build an inventory; but whining about it in NY Times articles is not going to get you that inventory.
Contractors are required by the Department Of Labor to be treated as less than employees, to avoid the Department Of Labor declaring that they are in fact employees.
balderdash - the DOL doesn't "REQUIRE" them to be treated as lesser than, it "ALLOWS" organizations to treat them as lesser than employees. And the DOL is trying to change that. Businesses take advantage of insufficiently stringent regulations.
They *REQURE* that they do that in order to not be screwed over by the DOL converting those contractors to employees when they are in fact contractors, and there is in fact a contractual relationship, and both parties have consented to the fact that the relationship is company/contractor rather than employer/employee.
Because you can shove unfunded mandates onto employers for having employees, but you can't shove them onto a company for having a contractor. Like health insurance, and workman's comp (which states borrow from as if they are the federal government borrowing from the social security fund: no intention of ever paying it back) income tax collection and social security tax collection. With a contractor, the state and the fed have to go afer each contractor individually: *OF COURSE* they prefer that the regulatory and enforcement burden be offloaded to employers, instead of having to pay for them themselves!
So let's change the wording from "REQUIRES" to "will take any opportunity to fuck them up the ass by converting the contractor's designation to employee at any opportunity"?
No. I am only counting the countries in the European Union. Nobody wants to live in Russia under Putin, and no one wants to live in a number of the other Eastern block countries which are technically part of continental Europe, but not part of the European Union.
In fact, there are several countries in the European Union itself which should probably not be counted either, since Germany is already fighting against (via maltreatment) of refugees from Turkey and several other countries where it's unpleasant to live due to, among other things, restricted communications access and Internet filtering in order to implement draconian social policies.
Counted that way, they have about 1.6X the population of the U.S. in about 1/3 the total area of the U.S. (less, on both, is you leave out the undesirable parts -- say about 1.3X the population in about 1/4 the area of the U.S.).
No, we are not going to relocate Silicon Valley to Bozeman, Montana, just to make Europeans happy about our average population density, or to make San Francisco more desirable.
We can tell you're full of shit when you mention China.
They have no need of your rice. One can only assume the rest of your post was just as idiotic.
We can tell you are full of shit when you post as an AC, and fail to cite statistics or source. Here, asshole: USDA information on China's rice imports.
"Rice imports by China are expected to set a new record in 2015, surpassing 2014 levels by 200,000 metric tons and marking the fourth consecutive year of record imports. Rice imports surged in 2012 to more than 7 times the average of the previous 5 years, and continued to grow each year thereafter. China remains the world’s largest rice producer and consumer, and has been largely self-sufficient in rice for more than 30 years and until recently, was typically a net rice exporter. In 2012, China surpassed Nigeria to become the world’s largest rice importer. "
China has been a net importer of rice since 1981, and is currently the words largest rice importer, as of 2012.
Building up is BS. It's not everyone's dream to live in Skyscrapers with no windows to open, no greenspace and no fresh air (even if there were windows to open).
No, you're right. It's just all of the people who want to live in San Francisco, Austin, Mountain View, Seattle, New York, and Chicago who can't build out... you know: where all the jobs are located.
You definitely CAN build out. Look at Europe. You can find IT work in a city of 150000 ppl and live in a real house, with a garden and other great things. And this repeats over and over again. The US is not Japan. There's space. Use it. Properly
How about Europe starts granting emigration visas? Then we can just send you our tire, our poor, and our huddled masses, yearning to be free, and live in sprawling metroplexes, where the primary benefits are a working public transportation system, because, you know, Europe is kind of tiny.
Slashdot Mirror
Which is irrelevant, since it's not that system that failed.
No, it's the system that would have NOT failed, had they been using it instead of the shiny, new, failure prone system.
And surgeons responsible for cutting open live human beings should be capable of not leaving tools in the person they're operating on, but it still happens. Professionals make mistakes. Garbage collection is a useful tool to make it more difficult to screw up.
If only there were some kind of portable machine one could use in order to look for metal left in a patients body...
If only there were a requirement to do things like count gauze pads before and after surgeries, and then account for any numerical discrepancy as being ON (not IN) the patient, and/or going into the biohazard disposal unit. You know: a documented procedure.
Oh, wait: there is.
Very funny guys.
If you can't handle explicit memory management, then you can't write in MISRA-C, and if you can't write in MISRA-C, you should probably be kept as far as possible away from coding on life support systems, because you are not going to be good at it.
The idea that the could "run out of memory" in the first place implies heap allocation, at the very least, which is prohibited in the MISRA-C standard.
You may remember the Toyota acceleration bug in the ECC; it was due to a practice which, had they been enforcing their own coding standards (which were a partial intersection with the MISRA-C coding standard, which was under development and nearing completion at the time), they would have caught the bug before it started ending up in dead people.
What is heretical about novel new genes arising out of junk DNA? Molecular biologists have known for many years that so-called "junk" DNA played a number of roles; regulatory, and that most certainly novel genes could arise.
There is a small percentage of biologists which really, really would like acquired traits to be heritable (as a work study job in college, I worked in the lab of one of them, and we were cautioned not to talk about his theory outside the lab).
Every so often, one of the proponents of the idea of heritable changes due to environmental pressure, or more formally, either Lysenkoism or Lamarckism, tries to find a mechanism that could make it work. Even though it's never been demonstrated (the biologist in the lab I worked at was attempting experiments with, among other things, chelodina longicolis diets, to force physical changes, which he hoped the offspring would inherit, even though not on that diet).
This theory is what's known as "soft inheritance".
The main premise for its development in the first place was that Joseph Stalin was all pissy about genetic being a non-Soviet idea, and wanted an nice, Soviet alternative that better fit the ideology he put forth. This actually influenced a lot of decisions in Soviet agriculture that didn't work out badly enough that they ended up importing wheat from the West.
The last go-round was trying to use introns as a mechanism whereby he introns were involved in making traits heritable (and before that, it was endogenous viruses, such as PERV - Porcine Endogenous Retro Virus). Those were the main ones. The've also tried to explain it with varying degrees of gene methylation, and so on. Todays flavor is non-coding DNA (the correct scientific name for "junk genes").
Unless the can demo it in plants, mice, or fruit flies, etc., don't expect that the idea will go anywhere directly.
The sad part is, if they had concentrated on the punctuated equalibria model, which the article mentions, instead of trying to explain it as a short scale inheritable phenomenon, the might have had a really great argument.
(Yes, I am in the 90% who are skeptical about this, without further evidence and perhaps a demo).
I kinda doubt that, My understanding is most of the US's air-traffic control systems (and software) is ancient .
Somehow, I doubt it was 2,000,000 lines of assembly language.
Actually, the old system was a bunch of 68K assembly. Nowhere near 2,000,000 lines of it. I know one of the guys who wrote some of it.
THIS.
So they "invented a new tool"? Nope.
Sorry to tell you, but they are 13 years late to the party.
A Foresight.org project on 2002 was the release of Ka-Ping Yee's CritSuite.
It was also "invented" independently as "Third Voice" in 2004.
It's cute that they believe they've done something unique with the grant dollars (that they should probably be actually spending on dealing with climate change, rather than beating reporters over the head with foam "pool noodles"), but what they've done is far from new.
Funny how Slashdot's fierce Fighters for Privacy turn a blind eye when it's something they don't care about that gets hacked.
I think we are just laughing our asses off about trying to decide whether Pat Robertson was actually able to employ competent system crackers, or whether the security on the site was so lame, given its purported purpose and clientele.
Also, note that this isn't a privacy issue, it's a database hack; it's just incidental that it involves a website that contained prurient details of illicit affairs, rather than, for example, credit card data for everyone who used a credit card at Target, Home Depot, and Lucky's, or the OPM's personnel database and security clearance validation records (I still think the goal was data insertion on that hack, rather than obtaining the data, or it was at least a dual goal).
If you think that information about sexual exploits, however obtained, should not be dumped en masse to the press, then you must be from Europe and it's "right to have your hookers dressed as Nazi photos be forgotten" laws. Because in the U.S. -- once it's out, it's out.
I'm waiting for the data mine on specific famous persons and political figures to be published with quite some interest... wonder if anyone is going to be dropping out of the presidential race over it...
"Jared Newman investigates why Flash won't disappear from the web anytime soon."
Because the media companies won't spend money on the alternative ..
That's actually not true.
They are willing to spend a crap-ton of money on the alternative. It's just neither of the two competing HTML standards camps is willing to implement their draconian DRM demands for them, or deal with their "commercial insertion problem", other than making it a server side problem, where it has to be placed into the content stream, and the user of the browser then has the option of skipping it by jumping ahead in the buffer.
No forced commercials means they will not be moving their content to HTML5 soon.
NB: One company was even demanding that they be able to implement "pause on mute" within the streaming protocol, so that not only did you have to *watch* the commercial to watch the next section of content, you had to *hear* it as well.
"IBM 'TrueNorth' Neuro-Synaptic Chip Promises Huge Changes -- Eventually"
So basically... the chips are thinking about it?
An eager explanation...
Most of the sites offering television content -- Hulu, NBC, etc. -- use a video player and content library management system from a company in Orem, Utah, in order to implement their commercial insertion mechanisms in order to monetize their online content. This player only works in flash, and simply would not work in HTML5, even if you wanted it to, without giving people the ability to block commercials from these sites.
Other companies content is managed, and the DRM protection mechanism, and operational ticketing mechanism, is implemented using Adobe Access. Adobe Access is a content management system based on an older product known as Adobe Flash Access (actually, it's the same product, renamed). The content control mechanism therefore requires the use of a flash player in order to implement the DRM. Companies using this content platform include Google (Google Play) and Amazon.
From a non-content perspective, a number of companies do a couple of things which are impossible to do in JavaScript, but which are often highly convenient as UI features, and flash allows a click bypass in the UI for these companies in order to implement their desired behavior. One of these is GMail, which has an upload drop-zone mechanism for email attachments (actually, lots of web mail applications utilize flash specifically for this feature). Another is the Amazon "one click purchase", which would otherwise be "two click purchase".
Finally, a lot of content is simply authored with Adobe Content Creator tools. Say what you will about flash, but Adobe has the finest web content authoring tools, content management system, and so on in the business. Others are at best a poor second, and none of them contains all of the capabilities of Adobe Creative. Sadly, that lets them, as part of their content creation, transparently insert flash into your web content. You can disable this (there are some obscure settings to allow it), but doing so disables some of the features you can use on your web sites. There are more modern methods of adding those features back in, but Adobe has not provided alternative implementations, since they feel "the flash implementation is good enough".
That explains everything that I'd be personally "reluctant to discuss" about why my site was continuing to require flash. As you can see, there are reasons, and they aren't great, but it's not like everyone is using flash to crap "super cookies"; there are other ways to track you.
Yet another article that assumes Beijing = China. Sigh. It's like there's only one city in China. Imagine if European journalists assumed New York City was all there was to know about the entire USA. And China is even bigger, and has four times the population! I think the problem is due to the fact that most Western journalists live in Beijing, and they are not really interested in reporting about anywhere else other than where they live.
It's pretty bad in most areas of China where there are actually monitoring stations (which is where there are actually people). Here's a pointer to an interactive map which demonstrates it graphically
http://aqicn.org/map/californi...
One has to wonder what the hell is going on in Kashi and diqu zhan Hotan, which are near the Kyrgystan border, and have the highest and second highest (respectively) "bad" numbers of any reporting stations in the world.
This is called closed-mindedness and provincialism if it occurs in rural people, but now it's suddenly acceptable?
Actually, it's called "journalists are assigned "minders" and are only permitted to go wherever the heck the government lets them go, and nowhere else, so they only see what the government allows them to see". Welcome to China; new employee orientation for the state controlled media for foreign journalists is on alternate Tuesdays.
When I lived in the Bay Area, there was a fire at a recycling plant that caused some reading to go 400, or 500 or something in Redwood City. It might have been particulates, not sure.
Yeah, that's a teensy bit down from Hiller. There was some suspicion of them not being able to handle the recycling load, and "accidentally" setting the materials on fire (the plant itself was untouched). Other theories included spontaneous combustion due to thermal rise during decomposition (only it was mostly paper).
It was particulates in the 76 or so today, due to smoke from the wild fires (which are actually pretty far away). Everyone got an emergency services robocall. Most places in China are about that, according to the monitoring mapping service (aqicn.org), but there are some that are running about double. Highest I saw was a 953 on the China/Kyrgystan border (kinda insane), and a couple real hotspots around Beijing.
I found it interesting that they shut down the San Francisco station (it must have been showing numbers that were unfavorable to San Francisco tourism). Worst in the U.S. is Medford Oregon; most other hot spots are in Washington State. There's a 229 in the Czech Republic. Russia has exactly one monitoring station; I'm going to guess it reports whatever Putin wants.
If you are interested in the world map (navigable Google Maps map), it's here (I left it centered on China):
http://aqicn.org/map/californi...
Thanks, insightful comment. I'd like to add little more about how Windows handles USB devices though, as it is somewhat relevant here.
Thanks; interesting tidbit about the Windows model; it makes sense, but as I said, I rarely see devices in the wild with actual unique serial numbers. Port topology is, as I said, also used by Apple, and you can be in trouble if you change it while the machine is asleep and later wake it up. Mac OS X has other issues with sleep/move/wake that can be pretty nasty; kind of off topic (not USB related at all), but they will all self-resolve in at most 3-5 minutes.
I've noticed that the FTDI driver can break USB serial number reporting under certain circumstances too. I'm still narrowing it down, but USB 3 ports seem to be worst affected for some reason.
The Windows way seems like a reasonable way to handle a bad situation.
FTDI devices are really problematic. On USB 3, they tend to trigger "downstream charging port mode", and USB 3 frequently has a hard time recovering from that (which is why when you plug a Samsung phone into a Samsung computer, it will often charge, but then you can't talk to it with the computer, until it's fully charged, or you replug it frequently. FTDI also has driver problems in that the driver can try to run if you unplug the device, and if you did a teardown, panic either Mac OS X or Linux machines. Both OS's have written their own non-vendor drivers to deal with the issue.
Typically, I just put the FTDI serial dongles at the very end of my port hierarchy/topology, and then they don't do the "problem child" thing and screw with your other devices.
So well you posted a lot of text, and many people think you are quite insightful, you do not actually tell us what the problem is (apart from referring to the independent keyboards issue which is /a completely different/ issue).
Linux has had, via udev and systemd persistent device nodes for years [...]
They are not persistent for pluggable devices for which you have not made specific configuration file changes to the udev.conf for the device.
In addition to the explicit conf changes, you may also have to make a udev.rules entry, such as:
KERNEL=="whatever", RUN+= "udevadm trigger --action=remove --sysname-match=%k"
as a pre-run rule to remove the old device so you don't end up with a new device node with a different name/minor replacing it. You have to explicitly know the name "whatever" of the device in question.
In Chrome OS we had to add a rule like this for the Synaptics Touchpad on the Alex (Samsung) Chromebooks so that the device would not disappear on upgrades, due to the timing of rebuilding the kernel cache.
So yes, I did tell you what's wrong; no, you are not correct about it being ACPI -- that's an unrelated Linux religious issue; no, this is not my first rodeo: I do know what I'm talking about here.
because he is actually wrong. He has some personal problem with the USB implementation I think. But I think he is either wrong, or fails to describe the problem properly. In any case, its just a minor issue, ACPI issues are a far greater source of pain with resume.
ACPI is not the problem with the controllers going out, unless you are claiming the reason the USB devices are not coming back is due to a failure of the ACPI code to repower the EHCI/UHCI components?
Yes, there are major suspend/resume problems around the use of ACPI power states to implement suspend/resume. I really don't care about those, since those are "do as I do, not as I say" problems; they are due to expecting the ACPI implementation to religiously adhere to the specification so that you can push the envelope, vs. "do what Microsoft Windows does, and don't push it".
Pragmatically speaking, expecting religious adherence to the ACPI specification, instead of just doing what Windows does, is a moral high ground that will get you technically correct code and a non-working system. FreeBSD does a lot better about getting this right, but is still not as good at it as Windows, since vendors only care about working with Windows.
This is similar to Linux being all religious about UEFI and ExitBootServices(); there's a static library portion which doesn't have a UEFI mapping in the Intel reference implementation (which *everyone* uses), and so you if you call that, and then don't leave the recovered memory alone anyway, there are certain EFI calls that, if you make them, even though they are supposed to be available runtime, you are screwed.
When we are looking at USB devices is it possible this could be handled later in the boot process by having all devices re-polled and connected as new devices? So on resume previously connected devices are simply treated as new devices. It would add a delay to the resume process and I'm guessing there may be device confusion if you have 2 controllers attached, ie player 1s controller becomes players 2s after a suspend resume. But would this not be a simple, but manky, solution to the problem?
If you are talking about the application programs closing and reopening the devices, then yes, the controllers could be handled that way. You would have to do it in every single application, however, for every single device which was capable of moving around that way, so it would include audio devices, cameras, game controllers, and so on.
Further, you'd have to have a mechanism for notification of the applications that they need to close and reopen (and reinitialize) their devices.
This really can't be done at a system level, because on resume the devices look like arriving new devices, and the nodes are in use by existing opens. So it is, for example, easy to "lose" a built-in camera or keyboard or audio device, because those end points are in use, and the "new" device gets assigned a new end point.
One of the things I fixed in the Chrome OS testing environment was that the cameras got "lost" during factory testing on a number of devices where they were interfaces via USB (but were built into the clam shell). This meant that the test harness, which set up devices up front, would lose the devices because there were "there and open" in the Python code, but "not used yet before the suspend/resume cycle".
The fix was to move the camera open/close code, ant to modify the code to search for specific attributes which meant "camera attached to this node", rather than "node for a camera".
Technically, suspend is not the problem. The problem is resume. It doesn't probe and reattach the controllers to the same point in the device tree that they were in when the system was suspended. Since those are the device nodes that SteamOS has open at the time of the suspend, and they route to The Noplace(tm), the controllers become unresponsive.
This is a general problem in the Linux device model, and you can see problem in the device model poke their heads up in various places.
For example, if you plug in two USB keyboards, what you are going to see when you hit the caps lock key on one of them is that the caps get locked for both of them, but the LED indicating that the caps lock is on is only lit on the original keyboard. This is great fun, when you see the LED state on both keyboards relative to the caps lock state, when you bounce the caps lock key on the second keyboard.
Presumably, this was implemented this way in order to allow multihead operation, with the keyboards separately (and explicitly -- which they are not by default -- to separately running console instances. But it's indicative of deeper problems in the model.
The reattach operation should (in theory) be handled by the udev mechanism, which I'm told is subsumed in systemd; however, they've faithfully reproduced the problems in the original implementation in the replacement as well (so it actually doesn't matter which one SteamOS is using, so please don't argue about that crap).
There are two implementations that do work; however, they in fact work differently.
The first implementation is in Windows; it works by directly assigning the device descriptor based on the USB vendor and device ID. What this effectively means is that when you resume, everything gets the same descriptor (or a replacement) that it had going own. The keyboard "problem" is handled in Windows by making them "the same device" -- in other words, you caps lock on one keyboard, it does it on both, and you undo it, and it does it on both.
The second implementation is Mac OS X; it's handled by explicit enumeration order of USB bus devices, and the using the USB vendor and device ID *and* the enumeration position ("bus ID) to uniquely identify the device.
You'll also notice, if you look, that Linux has problems with keyboard internationalization and locale. This is most easily seen by using a locale specific keyboard, and having it not be recognized. Further, you'll notice that the character set differences are handled by tables in X and Weyland, and these are not the same as the console tables (i.e. the USB ID of the keys is not propagated through the full input stack, and there's a difference in operation between sending the events up through the console, vs. sending events up through the X Server). To fix this would require moving the HID key value translations into the kernel keyboard driver, rather than having it (mostly) in user space in all three instances (X, Weyland, console).
Finally, Apple is one of the few vendors that actually correctly fills out the USB device serial number field correctly, so it's hard to use that as a unique identifier (specifically, it makes it really hard to mask program all your controller chips, without adding a "burn the fusable links" step).
Further, they are also one of the few vendors that sets the locale field in their USB devices (most laptop vendors will get this right during manufacturing, by placing the value into the BIOS, since the laptop keyboard is actually matrix decoded by the EC via a grid hooked up to GPIO pins, and then the EC pretends it's an 8051 with a PS/2 interface for the keyboard to mux a PS/2 trackpad -- e.g. like a Synaptics -- to look like a standard PS/2 keyboard and mouse).
Apple handles for unrecognized keyboards by having you press "the key right of the left shift key" and "the key to the left of the right shift key", and then uses the key IDs. Not ure if they have a patent on it, but it's a lot more clever than what Linux or the BSDs do to
Reach your data cap up to 40X faster!
That is all.
How much reading and writing is necessary for an average person to succeed. Donal Trump speaks at an elementary school level.
This is actually a valuable political and business skill. It's called "being able to speak to your audience", and it's fundamental to being able to communicate your point to people at that skill level themselves. Reagan had this same skill, as did Kennedy. You will find that all the people we consider "great orators" had this skill. Jimmy Carter, to a lesser degree (or he would have had a second term); surely people remember his "I want to be your president, because I like you". Mr. Rogers level dialogue.
I personally found both Ross Perot, and later, Steve Forbes, both intelligent and insightful about the problems we face as a nation, and they laid out their plans in great detail. But they both talked well over most of their audience's heads, which rendered them complete unelectable.
I don't know anyone in the Democratic field at this point who has a similar skill set to Trump. Perhaps Bernie Sanders. Definitely not Hilary Clinton.
Do not underestimate the value of speaking to your audience.
The article is inane.
First, there is no such thing as a "rock star front-end developer". The front-end space is actually constrained enough that it's possible to know all of it, and act within the limits of those constraints all the way up to the boundaries and no further. So it's a pretty rote work position for a developer.
Second, their definition of what constitutes a "rock star" is inane, in that it's a stereotype of the behaviours of people who have achieved "rock stardom", and not a description of their capabilities. The issue is one of capability, and a "rock star" musician, like it or not, is capable of doing things which others are not capable of doing. It's rather as simple as that, and hence the migration of the analogy into other areas of human endeavor.
Having a developer who is a "rock star" is a significant competitive advantage, in that they will be capable of doing things which others are not capable of doing. This is a competitive advantage, in that it acts as a barrier to entry to your market, because it means that your competitor simply can not hire someone that is capable of competing on your level no matter how much they pay, short of hiring your "rock star" away from you.
This is the equivalent of "first mover advantage", without the pressure of having to execute quickly in order to maintain that advantage. You don't have to spend crap-tons of money on "ramp speed" and "burn rate" and "time to market", and "runway"; as a result, the VC feeding frenzy ends up owning less of a chunk of your company, and you get to be rich instead of making *them* rich (or while *also* making them rich, but not as rich).
So yes, it pays to hire "rock stars" for strategic things (and again, front end web design is strategic, but from end web coding is not; designers can be "rock stars", but front-end developers can not).
So yeah: there are some stupid categories in which to advertise that the person you hire be a "rock star"; there are many others where it's not actually silly, it's smart, or it's even imperative to have one.
Yes: it's really frustrating to recruiters, and to job sites like DICE, when someone asks for a "rock star", and they are incapable of delivering one to their customer, because they have none in inventory. Lump it. Be more desirable to "rock stars", and you will be able to build an inventory; but whining about it in NY Times articles is not going to get you that inventory.
Contractors are required by the Department Of Labor to be treated as less than employees, to avoid the Department Of Labor declaring that they are in fact employees.
balderdash - the DOL doesn't "REQUIRE" them to be treated as lesser than, it "ALLOWS" organizations to treat them as lesser than employees. And the DOL is trying to change that. Businesses take advantage of insufficiently stringent regulations.
They *REQURE* that they do that in order to not be screwed over by the DOL converting those contractors to employees when they are in fact contractors, and there is in fact a contractual relationship, and both parties have consented to the fact that the relationship is company/contractor rather than employer/employee.
Because you can shove unfunded mandates onto employers for having employees, but you can't shove them onto a company for having a contractor. Like health insurance, and workman's comp (which states borrow from as if they are the federal government borrowing from the social security fund: no intention of ever paying it back) income tax collection and social security tax collection. With a contractor, the state and the fed have to go afer each contractor individually: *OF COURSE* they prefer that the regulatory and enforcement burden be offloaded to employers, instead of having to pay for them themselves!
So let's change the wording from "REQUIRES" to "will take any opportunity to fuck them up the ass by converting the contractor's designation to employee at any opportunity"?
Tiny? Wow. Are you including Eastern Europe?
No. I am only counting the countries in the European Union. Nobody wants to live in Russia under Putin, and no one wants to live in a number of the other Eastern block countries which are technically part of continental Europe, but not part of the European Union.
In fact, there are several countries in the European Union itself which should probably not be counted either, since Germany is already fighting against (via maltreatment) of refugees from Turkey and several other countries where it's unpleasant to live due to, among other things, restricted communications access and Internet filtering in order to implement draconian social policies.
Counted that way, they have about 1.6X the population of the U.S. in about 1/3 the total area of the U.S. (less, on both, is you leave out the undesirable parts -- say about 1.3X the population in about 1/4 the area of the U.S.).
No, we are not going to relocate Silicon Valley to Bozeman, Montana, just to make Europeans happy about our average population density, or to make San Francisco more desirable.
We can tell you're full of shit when you mention China.
They have no need of your rice. One can only assume the rest of your post was just as idiotic.
We can tell you are full of shit when you post as an AC, and fail to cite statistics or source. Here, asshole: USDA information on China's rice imports.
http://ers.usda.gov/data-produ...
"Rice imports by China are expected to set a new record in 2015, surpassing 2014 levels by 200,000 metric tons and marking the fourth consecutive year of record imports. Rice imports surged in 2012 to more than 7 times the average of the previous 5 years, and continued to grow each year thereafter. China remains the world’s largest rice producer and consumer, and has been largely self-sufficient in rice for more than 30 years and until recently, was typically a net rice exporter. In 2012, China surpassed Nigeria to become the world’s largest rice importer. "
China has been a net importer of rice since 1981, and is currently the words largest rice importer, as of 2012.
But you know, don't let facts stop you.
Building up is BS. It's not everyone's dream to live in Skyscrapers with no windows to open, no greenspace and no fresh air (even if there were windows to open).
No, you're right. It's just all of the people who want to live in San Francisco, Austin, Mountain View, Seattle, New York, and Chicago who can't build out... you know: where all the jobs are located.
You definitely CAN build out. Look at Europe. You can find IT work in a city of 150000 ppl and live in a real house, with a garden and other great things. And this repeats over and over again. The US is not Japan. There's space. Use it. Properly
How about Europe starts granting emigration visas? Then we can just send you our tire, our poor, and our huddled masses, yearning to be free, and live in sprawling metroplexes, where the primary benefits are a working public transportation system, because, you know, Europe is kind of tiny.