I know Oakland University used to offer the same class as either a masters level (5xx class) or a undergrad level (4xx class). The expectations for the masters level students are higher, but the lectures are the same. Since they cater their masters classes to working professionals all of the classes were 6pm or later starts.
UIUC is similar. Most masters-level courses can also be taken as an undergraduate with a lighter workload (e.g. less assignments, or "solve 4 out of 5 problems" on the same assignments). The class hours are not tailored for working professionals, but everything is available online as a live stream or a recording later. I don't remember how the live streaming worked, but I remember hearing an occasional question from a remote adult student, so it was still interactive if you could work out flex-time from your day job.
That's crazy. With the way that HR plays buzzword bingo before passing resumes to hiring managers, anyone who has those certificates would be absolutely nuts to not list them.
That depends on where you work. In my office, HR generally does not play buzzword bingo, and there are very few relevant certifications which could be listed (i.e. Cisco certs for a network admin position). So if somebody lists a bunch of (IMO) irrelevant certifications, it is probably a sign that they don't understand the job they are applying for.
While I'm not an employer, I have been active in a lot of technical interviews and hiring decisions. So far I've regretted about 1/15 of my own recommendations, which I think is a decent success rate.
1. I do look for a technical degree though not necessarily a CS degree. There's a certain maturity of the thinking process that rarely happens outside of college.
I have worked with a handful of software devs who do not have degrees. One of them was active on F/OSS mailing lists preceding my birth. The other one easily has a more thorough and analytical approach to problems than most other people. For both of these guys, the lack of a degree would have been an issue in initial resume filtering, but a total non-issue as soon as the technical interview started.
2. University of Phoenix, DeVry, Strayer and similar "degree mills" do carry a negative stigma.
3. Certifications can be very bad. If you have one or two very strong certifications, like CCIE, they'll help you... On the other hand, if you have 10 weak certifications (CCNA, MCSE, A+, Security+, etc.) and you list them all, that's a big negative.
Agreed on both accounts. I've gotten in arguments with friends in IT about this. I think the certifications and weaker colleges are fine for positions which are generally more reactive in nature, and constrained by someone else's user interface. In such an instance, it's all about optimal tool usage to keep the gears turning. For a creative role (e.g. semi-autonomous developer or IT architect), I think the presence of low-value certifications indicates a sort of lazy approach.
Have you looked at how the program at your alma mater has grown since you graduated? Last time I chatted with CMU students (at an internship), they had entire sub-tracks within the CS department for software engineering, human factors, and other topics more relevant to the art of computer programming, as opposed to pure math and algorithms on a turing machine. A quick DDG shows the Software Engineering Institute which IIRC was related.
Other posters have mentioned the continuing existence of generic block boxes. There is also the Creator series, where every kit comes with multiple instructions, and the vast majority of pieces are generic. I, too, was disenfranchised with Lego's theme sets over the last decade; but IMHO Creator recaptures the flexibility of the "golden oldies".
The great thing about web browsers and the nature of CSS is that it is typically easy to disable garish styles on well-mannered websites (those built simply or with accessibility in mind):
Opera: Page > Style > User Mode. Easy peasy.
Firefox: View > Page Style > No Style. Seems to require "Menu Bar" to be visible, which is annoying.
IE: View > Style > No Style. But wait, this seems to do nothing! Oh well... Alt+F4.
Guns have no use but to Kill... No one has any use for a gun
Regardless of the 2nd Amendment, I have a legal right to kill someone on my property if I believe my own life is in immediate danger or if they are about to commit a felony (yay castle doctrine). Personal self-defense training, marksmanship, and gun ownership don't grant me that right, but they sure make it easier to execute.
They could sue me for all they want and would never see a dime of any award for their hospital bills and pain and suffering.
The problem is that this seems to be a major source of income for them. So, if anyone wishes to do violence against them, their past behavior actually suggests a scorched-earth approach: don't touch them at all, or kill the entire family. Unpleasant either way.
I do think the funerals are very different from the other places that I know contain "free speech [sic] zones": airports and political events. A funeral is a very private ritual, and participants are expected to be in a compromised mental and emotional state. I'm not exactly sure how to codify that in a legal fashion, but I know the difference is there.
How exactly would you "dis-allow" this sort of free speech?
IANAL, but I suspend some sort of preemptive restraining order would work: Members of WBS not to be within X distance of any of the victims' funerals (or processions while mobile). I don't think it creates any slippery slopes: since the WBS has already broadcast their intentions, there is no speculation required for the scope of the restraining order.
I see it as a test for those who claim to be champions of the freedom of expression.
Funeral rituals go back much farther than any concepts of freedom of expression. I would not be surprised to see many otherwise-champions choose "honor the deceased" over "freedom of expression" for this specific set of deceased individuals. I do hope the Patriot Guard Riders keep them corralled.
Where do these x10 frames come into play? In dejuddering. The panel interpolates frames so the motion is smoother.
Even if you have motion-estimation algorithms turned off, the higher output refresh rate can help for a few reasons:
Higher refresh rate means faster color changes, which means less time "in between" one frame and the next, which means more time is spent showing the desired image.
Some people are sensitive to the green phosphors on plasma TVs, and they see ghosting. This ghosting should be reduced with a faster panel.
Some people are sensitive to low refresh rates, period.
Refresh rates which are divisible by both 24 and 60 allow the TV to detect and undo 3:2 pulldown for material which was originally shot at 24fps and the distributed as a 60fps DVD
I hate motion estimation myself, I find the errors to be disproportionately distracting. But the higher refresh rate is still valuable.
...the transition from incandescent bulbs to compact fluorescent... strikes me as nostalgia for a (mostly) inferior product.
CF lighting uses more hazardous materials (must recycle, cannot go into trash), generates poorer-quality light for the consumer price-point (CRI ~80 vs. 95+), and is more sensitive to a dirty power grid (ballast). The tradeoff for CFLs is that you're sacrificing all-around quality for energy efficiency, and kvetching over this transition is completely warranted. It's not nostalgia for an inferior product at all.
I did not say that the total range is noticeable. I said that 2deg F is noticeable in that range. We are more sensitive to temperature variation than you are giving credit for.
The difference between 71F and 73F is not exactly something a human is tuned to.
Having a home with poor air circulation and a thermostat which runs warm as a consequence, I have to disagree: in the winter, a few degrees Fahrenheit in the 66-72 range are definitely noticeable.
... some HFT systems are given a 30ms "preview" of orders that are yet to be acted upon...
That behavior was called "flash orders", and it seems to be the most frequently cited evidence of "front-running" in the public markets*. I've already outlined the 3-month timeline elsewhere in this thread. They have long since died, but of course the articles are perma-linked with no indication of updates or new developments on the topic.
*Real front-running requires knowledge of someone else's orders before they hit the exchange, in order to be risk-free. This is impossible for proprietary traders, and is illegal (but theoretically possible, cough cough) for a brokerage with its own prop desk. But flash orders were pretty darned close, because the information was deliberately delayed going to CQS.
Here's the thing: is microwave faster than fiber optic for latency?
Yes. Speed of light is slower in fiber optic cable than air at STP.
No.
Oh whoops! Well, to humor you, there is a technicality: Like any other networking gear, microwave dishes will have a latency for routing the information to/from fiber-optic. If that latency outweighs the speed-of-light gains for the given distance, then the microwave link is worse. However, modern microwave gear would have no reason to exist if the engineers hadn't solved this problem.
For anyone who doesn't have a login for ft.com, you can view the google cache (High Frequency Trading Under Scrutiny). The specific practice dubbed "flash orders" is not described in further detail, but here is a timeline of the relevant press releases:
Before June 3 (sorry, cannot find any links): BATS,NASDAQ propose "flash orders" to SEC, (weakly) justifying compliance with existing regulations
So yeah, the particular article you linked is abso-darned-lutely correct about flash orders. But it's wayyy out of date. If you read through the various exchanges' discussions and comments, there's some very interesting back-and-forth going on:
DirectEdge accuses NYSE of being anti-competitive: NYSE did not implement flash order types, and it was expected that these orders would shift liquidity to NASDAQ/BATS/DirectEdge. All exchanges acted in their own self-interest here, because NYSE is the figurative gorilla in the room.
BATS implements "me too" functionality to keep in competition with NASDAQ, but is very quick to distance itself from the controversy.
Several exchanges highlight the (historically) new trend of liquidity moving into dark pools, and the risk which that poses to price formation in the normal exchanges. Hey, looks like they were right!
Everyone releases news in lock-step. It's like a big game of chicken, nobody wants to publish early because it just gives "ammo" to the others.
garbage collected/automatic reference counting language
In Python (ref-counted GC), the file object would be destroyed as soon as it went out of scope, which would release resources back to the OS. In the last version of Java I looked at (a while ago), it would wait until the next GC event to free the resources. That's bad, because it introduces the possibility of nondeterministic OS errors. Open too many files before the GC activates and *boom* you hit your per-process file limit, or the kernel runs out of memory and starts OOM-killing, or somesuch. Unless Java has changed, relying on a destructor ("finalizer"?) to release an OS resource is a Bad Idea.
Yes, your example is a good demonstration of how not to use destructors and RAII. The same "disappearing filesystem" problem holds true for the C API: many people don't think about checking the return code of fclose, but it can fail with exactly the same errors as fflush if there is still any data buffered. A seasoned C++ developer will understand that because a destructor must be fail-proof, operations with mandatory side-effects (e.g. I/O) should not be relied upon within them.
Personally, I find destructors quite useful for a number of tasks:
Closing input files
Closing socket file descriptors (note: for a TCP connection, I would still call shutdown() first... I have been burned by file descriptors cloned on a fork before, and close() only decrements the refcount whereas shutdown() actually sends a FIN)
Decrementing an external refcount
Releasing a held mutex
Releasing an advisory lock on a file
I can't blame you if you haven't been exposed to these tasks, as they are heavily dependent on the type of application being developed. But maybe this will help you to see why we do use RAII and not MAII.
Slashdot Mirror
UIUC is similar. Most masters-level courses can also be taken as an undergraduate with a lighter workload (e.g. less assignments, or "solve 4 out of 5 problems" on the same assignments). The class hours are not tailored for working professionals, but everything is available online as a live stream or a recording later. I don't remember how the live streaming worked, but I remember hearing an occasional question from a remote adult student, so it was still interactive if you could work out flex-time from your day job.
That depends on where you work. In my office, HR generally does not play buzzword bingo, and there are very few relevant certifications which could be listed (i.e. Cisco certs for a network admin position). So if somebody lists a bunch of (IMO) irrelevant certifications, it is probably a sign that they don't understand the job they are applying for.
I have worked with a handful of software devs who do not have degrees. One of them was active on F/OSS mailing lists preceding my birth. The other one easily has a more thorough and analytical approach to problems than most other people. For both of these guys, the lack of a degree would have been an issue in initial resume filtering, but a total non-issue as soon as the technical interview started.
Agreed on both accounts. I've gotten in arguments with friends in IT about this. I think the certifications and weaker colleges are fine for positions which are generally more reactive in nature, and constrained by someone else's user interface. In such an instance, it's all about optimal tool usage to keep the gears turning. For a creative role (e.g. semi-autonomous developer or IT architect), I think the presence of low-value certifications indicates a sort of lazy approach.
Have you looked at how the program at your alma mater has grown since you graduated? Last time I chatted with CMU students (at an internship), they had entire sub-tracks within the CS department for software engineering, human factors, and other topics more relevant to the art of computer programming, as opposed to pure math and algorithms on a turing machine. A quick DDG shows the Software Engineering Institute which IIRC was related.
NASDAQ offers high-speed connectivity in Cartaret, NJ and a backup in Ashburn, VA; neither of which is next door to Times Square.
Other posters have mentioned the continuing existence of generic block boxes. There is also the Creator series, where every kit comes with multiple instructions, and the vast majority of pieces are generic. I, too, was disenfranchised with Lego's theme sets over the last decade; but IMHO Creator recaptures the flexibility of the "golden oldies".
FTFY. There are certain classes of modules which can be tested with 100% coverage of code paths and input data ranges, but not all.
The great thing about web browsers and the nature of CSS is that it is typically easy to disable garish styles on well-mannered websites (those built simply or with accessibility in mind):
Regardless of the 2nd Amendment, I have a legal right to kill someone on my property if I believe my own life is in immediate danger or if they are about to commit a felony (yay castle doctrine). Personal self-defense training, marksmanship, and gun ownership don't grant me that right, but they sure make it easier to execute.
The problem is that this seems to be a major source of income for them. So, if anyone wishes to do violence against them, their past behavior actually suggests a scorched-earth approach: don't touch them at all, or kill the entire family. Unpleasant either way.
I do think the funerals are very different from the other places that I know contain "free speech [sic] zones": airports and political events. A funeral is a very private ritual, and participants are expected to be in a compromised mental and emotional state. I'm not exactly sure how to codify that in a legal fashion, but I know the difference is there.
IANAL, but I suspend some sort of preemptive restraining order would work: Members of WBS not to be within X distance of any of the victims' funerals (or processions while mobile). I don't think it creates any slippery slopes: since the WBS has already broadcast their intentions, there is no speculation required for the scope of the restraining order.
Funeral rituals go back much farther than any concepts of freedom of expression. I would not be surprised to see many otherwise-champions choose "honor the deceased" over "freedom of expression" for this specific set of deceased individuals. I do hope the Patriot Guard Riders keep them corralled.
Even if you have motion-estimation algorithms turned off, the higher output refresh rate can help for a few reasons:
I hate motion estimation myself, I find the errors to be disproportionately distracting. But the higher refresh rate is still valuable.
CF lighting uses more hazardous materials (must recycle, cannot go into trash), generates poorer-quality light for the consumer price-point (CRI ~80 vs. 95+), and is more sensitive to a dirty power grid (ballast). The tradeoff for CFLs is that you're sacrificing all-around quality for energy efficiency, and kvetching over this transition is completely warranted. It's not nostalgia for an inferior product at all.
I did not say that the total range is noticeable. I said that 2deg F is noticeable in that range. We are more sensitive to temperature variation than you are giving credit for.
Having a home with poor air circulation and a thermostat which runs warm as a consequence, I have to disagree: in the winter, a few degrees Fahrenheit in the 66-72 range are definitely noticeable.
Wow, the FCC is really strict about their key signature!. Seriously, that's how I parsed the sentence on first read; I had to do a double-take.
That behavior was called "flash orders", and it seems to be the most frequently cited evidence of "front-running" in the public markets*. I've already outlined the 3-month timeline elsewhere in this thread. They have long since died, but of course the articles are perma-linked with no indication of updates or new developments on the topic.
*Real front-running requires knowledge of someone else's orders before they hit the exchange, in order to be risk-free. This is impossible for proprietary traders, and is illegal (but theoretically possible, cough cough) for a brokerage with its own prop desk. But flash orders were pretty darned close, because the information was deliberately delayed going to CQS.
Yes. Speed of light is slower in fiber optic cable than air at STP.
Oh whoops! Well, to humor you, there is a technicality: Like any other networking gear, microwave dishes will have a latency for routing the information to/from fiber-optic. If that latency outweighs the speed-of-light gains for the given distance, then the microwave link is worse. However, modern microwave gear would have no reason to exist if the engineers hadn't solved this problem.
R is commonly used for statistical number-crunching. On this, your memory sounds accurate.
So yeah, the particular article you linked is abso-darned-lutely correct about flash orders. But it's wayyy out of date. If you read through the various exchanges' discussions and comments, there's some very interesting back-and-forth going on:
Can Chuck Norris please do something about the useless backtraces generated by my optimized builds?
In Python (ref-counted GC), the file object would be destroyed as soon as it went out of scope, which would release resources back to the OS. In the last version of Java I looked at (a while ago), it would wait until the next GC event to free the resources. That's bad, because it introduces the possibility of nondeterministic OS errors. Open too many files before the GC activates and *boom* you hit your per-process file limit, or the kernel runs out of memory and starts OOM-killing, or somesuch. Unless Java has changed, relying on a destructor ("finalizer"?) to release an OS resource is a Bad Idea.
Yes, your example is a good demonstration of how not to use destructors and RAII. The same "disappearing filesystem" problem holds true for the C API: many people don't think about checking the return code of fclose, but it can fail with exactly the same errors as fflush if there is still any data buffered. A seasoned C++ developer will understand that because a destructor must be fail-proof, operations with mandatory side-effects (e.g. I/O) should not be relied upon within them.
Personally, I find destructors quite useful for a number of tasks:
I can't blame you if you haven't been exposed to these tasks, as they are heavily dependent on the type of application being developed. But maybe this will help you to see why we do use RAII and not MAII.