Tor seems to have 2 fundamental problems they can't fix:
1. Tor hidden servers seem to have a new "oops, totally not anonymous" exploit every few months since the dawb of the project.
2. No one one can seem to write a secure web browser. Even with js etc disabled, FF (and everyone else) has had flaws simply in the code that renders a page. Not common at all, but all it takes is 1, and nations can afford to buy such exploits (and we know they do).
There no actual evidence of a successful attack on the onion routing element of TOR, and the team seems do an OK job of fixing theoretical issues, but if servers are flawed and the browser is flawed, so what?
I think TOR is doing a great job, however, of protecting privacy from the likes of Google and Facebook. If that's your goal, TOR is a great tool.
There's a lot of case law on "boilerplate contracts". Leases have been far worse that EULAs forever, and all sorts of crazy shit has been tried, and then tried. A long contract with a layman that has some hidden abuse is nearly impossible to enforce, if such contracts are common without the abuse. However, a "normal" contract is perfectly enforceable, even if no one ever reads it.
Supply and demand will sort it out in the long run. What's terrible is when only new businesses get built, and no new housing - then it goes through the roof. That was happening for a while in Seattle.
My real hope is that all the new downtown housing drains the suburbs a bit, making the commute easier and suburban rents cheaper as demand declines.
I looked through the flat list of "stuff on sale" and could only wonder "who buys this shit in the first place". Useless. However, I thought to check I bunch of stuff I'm planning to buy in the next couple of months (new PC build) and discovered the power supply I had in mind had quite a nice discount -- significantly less than anything on pcpartpicker -- so I picked that up and saved $50 or so.
I think that's the recipe: have a wishlist ahead of time, and just check the list for discounts. If you're planning to buy anyway, any discounts you find are gravy.
Redmond and Bellevue aren't cheap by non-SV standards (I pay north of $2400 for a two bedroom) and the commute is getting measurably longer. Hopefully that will settle down once all this new housing comes online.
And here we have a trade group for the software industry essentially claiming the same thing. Coincidence? You decide.
Every place I've worked on the West Coast has a developer hiring bar that turns away between 60% and 80% of those they interview. "Experience" on the resume doesn't matter once you're that far - it's all about demonstrating coding skills.
I don't think it's a great system. I see too many candidate rejected who would be perfectly acceptable in the role. The bar is set high enough that normal random performance variation of the candidates mean there's IMO a 50% chance that any given qualified candidate gets hired.
When these companies complain about "lack of talent", it's nothing to do with H1-Bs, and everything to do with a misplaced drive to hire "only the best". It's good to filter out the losers who just suck up everyone's time on the team, but what I see happening is "only the lucky, among the good", which is just a bad system.
Still, a remarkable chunk of candidates with great resumes (at least a third) just can't code for shit, so I can't object to the basic idea of demonstrating coding skills in the interview.
That's actually cheaper than the the tech area of downtown Seattle, where a small studio apartment runs ~$1800 if you're moving in now. Of course, there are also 50-somehting new highrises going up in Seattle, mostly residential, so maybe that will help a bit.
San Jose is high, but it's not crazy. I found the pay more than offset the cost - just stay away from SF. Seattle is bordering on crazy, but at least they're building out.
You're "stealing" the money you didn't pay. The word makes sense. Taking-without-paying is stealing. Just as it makes sense to call it "stealing" if you don't pay for a service you consume. Not all theft is equal of course - the losses due to piracy are pretty theoretical, and I don't care much about them, but some money is certainly lost to it.
Only racists care as much about race as you obviously do. It you peer deep into any discussion looking for any way it could possibly be about race, then (a) you are a racist, and (b) the discussion is better off without you.
Google has at least made some minimal progress towards making the system less abusive to content creators. From what I hear, you can actually get your money back if you counterclaim successfully.
What Google needs to add is a more automated system for movie and game reviewers to counterclaim. Being a reviewer isn't a "get out of copyright free" card, and it's not like YouTube can ignore the claims, but they can make it easy for e.g. a movie reviewer to counterclaim fair use by checking a couple of boxes.
What we as a nation can do is fix copyright law for YouTube-style derivative works. Some sort of mandatory FRAND licensing to protect everything from people uploading "lyrics on screen" song videos, to internet radio stations. Keep the licensing costs legally to a point where the artists gets the ad revenue, but that's it, no takedowns, and public radio stations can reasonably afford the licenses they need for an internet "station".
Primitives don't throw exceptions. That's the whole point. (Streams are useless, not destructors.) fclose() doesn't have the concept of exceptions, nor do any of the C system libraries in any platform I've seen.
Destructors for any non-trivial class should be empty (unless you have some debug logging etc). For a non-trivial class, the members will clean themselves up if needed, even if you throw from the constructor. No destructor needed.
Don't put cleanup logic in some elaborate File class - like you say, that's useless. Put it in the file handle class, which is only a wrapper for FILE* or HANDLE or whatever on the platform.
C++ lets you have classes that clean themselves up. The useful pattern is to do this for primitives, not complicated objects. E.g
FileHandle file = fopen(...)
You use file just like you would a FILE*, except you don't need to explicitly call fclose(). No clean-up block at the bottom, you just return, and fclose(file) is called as part of exiting scope. You can call fclose() if you want to for some reason, and that's safe too. It's as idiot-proof as code is going to get.
That means you can return from any point in the code, and each such object you created will be unwound safely in the reverse order.
It's really a huge win, and the pattern is easily adaptable to anything with reserve-release bookends.
There is no problem that one can solve with object oriented design that cannot also be solved with structs and function pointers.
There is no problem that can be solved with structs and function pointers than cannot also be solved with raw assembly code, written directly to disk using a sector editor (or butterflies). But it's fucking stupid to choose such a non-productive way to code if you're doing it for a living (I recommend everyone does it once for fun, though).
The *only* advantage to object oriented design is more intuitive code organization, and that only within certain problem spaces.
Even if you never use inheritance or polymorphism (which certainly only makes sense for certain sorts of problems), putting your functions into the same namespace as the data they operate on is always going to make for cleaner code.
That right there is a significant win for C++.
For almost all kinds of problems, the fact that C++ destruction can automatically release resources, so you can safely return from anywhere in a function instead of having goto fail all over the place is a significant win for C++. Too many programmers are just hopeless at getting the "clean up everything before you exit the function" right 100% of the time.
For code organization, sometimes C++ "concepts" (static polymorphism, so no function pointers) are a nice win. Questionable whether the many lines of code you can save justifies the fact that you're writing template code, however.
The more you look, the more places you'll find clean little wins for C++, with no performance cost.
All change involves risk. You cannot change for the better without some risk. Some ideas are riskier than others, of course, but "actually control immigration for one" seems as conservative as a change can be.
Which only goes to show how deeply, thoroughly corrupt the US has become. We are no longer a nation of laws, but instead have one America for the powerful and celebrities, and another America for the rest of us.
The fact is that a tiny percentage of the population holds the vast majority of the wealth, and they are not spending it.
Wealth is not money; money is not wealth. If you had an even share of all the stock in all the publicly held corporations in the US, it would pay about $1k/year in dividends. Yay? You can't "spend" wealth. You can sell it, at which point you no longer have wealth, and presumably as everyone starts doing that it ends up concentrated in the hands of just a few people again (though it might take a while).
The fact is, the majority of Americans are stockholders, and the median American retires with a reasonable chunk of wealth that she then sells off gradually to live on during that retirement.
Maybe you meant something other than "wealth"? But I can't guess what that would be.
Customer satisfaction is an important goal for any company, but Nintendo especially. High server load for a game during "launch week" requires no ESP - it's entirely predictable. You don't want to build out to a scale that handles launch week load and keep paying for that forever - it can be triple your ongoing load.
If you have an "internal cloud", then you can do this all in your datacenter: provision a ton of capacity, then release it back to your pool for the next game to use. That can be economical. If you don't, then you can borrow capacity from one of the big cloud providers, just pay by the hour until you don't need it.
I do this sort of dynamic provisioning using the cloud for a living these days. It's really not that hard, if your service was built for scale to begin with, and it's quite inexpensive compared to buying servers that will be idle for a year until the next major expansion or whatever.
My current gaming rig is 6 years old. I've switched video cards twice, and I expect a couple more years of use out of it. It's still fast enough, since the vid card does most of the work.
Building your first PC is hard, sure. If you don't enjoy that sort of thing, buy an Alienware. PCs just don't "age out" the way they used to these days: it's no longer something you have to do every 2-3 years.
Are you also going to rent 3x the support staff for a week?
You're not going to have more support issues by provisioning adequate capacity.
Or maybe they just don't like to run their brand new intellectual property on someone else's servers, without an airtight contract, liability for the cloud company, the possibility to audit the provider, etc, etc?
Almost everyone new runs mostly on the cloud these days. Neither Amazon nor Microsoft is going to steal your IP. The whole point of the cloud is to make it trivial to provision large amounts of capacity, temporarily if needed.
Just rent 3x the servers for launch week. There's a whole "cloud" of such servers available. But perhaps innovations from the past 10 years are too new for AAA games.
Can't wait to see kids running around the slums, whore houses, drug dens, and all kinds of shitholes looking for stupid pokemons.
I've been amused by considering the special Pokemon you can only catch in the sewers, or on the grounds of a nuclear power station, or on a military base, or in a burning building, or on a construction site. But sure, crack houses are fun too.
The news is "The problem is that Pokemon Go is not officially available in every region". Shocking that AAA game companies still don't get the single most important fact about selling games. Seriously, WTF?
Slashdot Mirror
Tor seems to have 2 fundamental problems they can't fix:
1. Tor hidden servers seem to have a new "oops, totally not anonymous" exploit every few months since the dawb of the project.
2. No one one can seem to write a secure web browser. Even with js etc disabled, FF (and everyone else) has had flaws simply in the code that renders a page. Not common at all, but all it takes is 1, and nations can afford to buy such exploits (and we know they do).
There no actual evidence of a successful attack on the onion routing element of TOR, and the team seems do an OK job of fixing theoretical issues, but if servers are flawed and the browser is flawed, so what?
I think TOR is doing a great job, however, of protecting privacy from the likes of Google and Facebook. If that's your goal, TOR is a great tool.
There's a lot of case law on "boilerplate contracts". Leases have been far worse that EULAs forever, and all sorts of crazy shit has been tried, and then tried. A long contract with a layman that has some hidden abuse is nearly impossible to enforce, if such contracts are common without the abuse. However, a "normal" contract is perfectly enforceable, even if no one ever reads it.
Supply and demand will sort it out in the long run. What's terrible is when only new businesses get built, and no new housing - then it goes through the roof. That was happening for a while in Seattle.
My real hope is that all the new downtown housing drains the suburbs a bit, making the commute easier and suburban rents cheaper as demand declines.
I looked through the flat list of "stuff on sale" and could only wonder "who buys this shit in the first place". Useless. However, I thought to check I bunch of stuff I'm planning to buy in the next couple of months (new PC build) and discovered the power supply I had in mind had quite a nice discount -- significantly less than anything on pcpartpicker -- so I picked that up and saved $50 or so.
I think that's the recipe: have a wishlist ahead of time, and just check the list for discounts. If you're planning to buy anyway, any discounts you find are gravy.
Surely that should be "enstrongulate his vocabulizer".
Redmond and Bellevue aren't cheap by non-SV standards (I pay north of $2400 for a two bedroom) and the commute is getting measurably longer. Hopefully that will settle down once all this new housing comes online.
And here we have a trade group for the software industry essentially claiming the same thing. Coincidence? You decide.
Every place I've worked on the West Coast has a developer hiring bar that turns away between 60% and 80% of those they interview. "Experience" on the resume doesn't matter once you're that far - it's all about demonstrating coding skills.
I don't think it's a great system. I see too many candidate rejected who would be perfectly acceptable in the role. The bar is set high enough that normal random performance variation of the candidates mean there's IMO a 50% chance that any given qualified candidate gets hired.
When these companies complain about "lack of talent", it's nothing to do with H1-Bs, and everything to do with a misplaced drive to hire "only the best". It's good to filter out the losers who just suck up everyone's time on the team, but what I see happening is "only the lucky, among the good", which is just a bad system.
Still, a remarkable chunk of candidates with great resumes (at least a third) just can't code for shit, so I can't object to the basic idea of demonstrating coding skills in the interview.
That's actually cheaper than the the tech area of downtown Seattle, where a small studio apartment runs ~$1800 if you're moving in now. Of course, there are also 50-somehting new highrises going up in Seattle, mostly residential, so maybe that will help a bit.
San Jose is high, but it's not crazy. I found the pay more than offset the cost - just stay away from SF. Seattle is bordering on crazy, but at least they're building out.
You're "stealing" the money you didn't pay. The word makes sense. Taking-without-paying is stealing. Just as it makes sense to call it "stealing" if you don't pay for a service you consume. Not all theft is equal of course - the losses due to piracy are pretty theoretical, and I don't care much about them, but some money is certainly lost to it.
Only racists care as much about race as you obviously do. It you peer deep into any discussion looking for any way it could possibly be about race, then (a) you are a racist, and (b) the discussion is better off without you.
Google has at least made some minimal progress towards making the system less abusive to content creators. From what I hear, you can actually get your money back if you counterclaim successfully.
What Google needs to add is a more automated system for movie and game reviewers to counterclaim. Being a reviewer isn't a "get out of copyright free" card, and it's not like YouTube can ignore the claims, but they can make it easy for e.g. a movie reviewer to counterclaim fair use by checking a couple of boxes.
What we as a nation can do is fix copyright law for YouTube-style derivative works. Some sort of mandatory FRAND licensing to protect everything from people uploading "lyrics on screen" song videos, to internet radio stations. Keep the licensing costs legally to a point where the artists gets the ad revenue, but that's it, no takedowns, and public radio stations can reasonably afford the licenses they need for an internet "station".
Primitives don't throw exceptions. That's the whole point. (Streams are useless, not destructors.) fclose() doesn't have the concept of exceptions, nor do any of the C system libraries in any platform I've seen.
Destructors for any non-trivial class should be empty (unless you have some debug logging etc). For a non-trivial class, the members will clean themselves up if needed, even if you throw from the constructor. No destructor needed.
Don't put cleanup logic in some elaborate File class - like you say, that's useless. Put it in the file handle class, which is only a wrapper for FILE* or HANDLE or whatever on the platform.
C++ lets you have classes that clean themselves up. The useful pattern is to do this for primitives, not complicated objects. E.g
FileHandle file = fopen(...)
You use file just like you would a FILE*, except you don't need to explicitly call fclose(). No clean-up block at the bottom, you just return, and fclose(file) is called as part of exiting scope. You can call fclose() if you want to for some reason, and that's safe too. It's as idiot-proof as code is going to get.
That means you can return from any point in the code, and each such object you created will be unwound safely in the reverse order.
It's really a huge win, and the pattern is easily adaptable to anything with reserve-release bookends.
There is no problem that one can solve with object oriented design that cannot also be solved with structs and function pointers.
There is no problem that can be solved with structs and function pointers than cannot also be solved with raw assembly code, written directly to disk using a sector editor (or butterflies). But it's fucking stupid to choose such a non-productive way to code if you're doing it for a living (I recommend everyone does it once for fun, though).
The *only* advantage to object oriented design is more intuitive code organization, and that only within certain problem spaces.
Even if you never use inheritance or polymorphism (which certainly only makes sense for certain sorts of problems), putting your functions into the same namespace as the data they operate on is always going to make for cleaner code.
That right there is a significant win for C++.
For almost all kinds of problems, the fact that C++ destruction can automatically release resources, so you can safely return from anywhere in a function instead of having goto fail all over the place is a significant win for C++. Too many programmers are just hopeless at getting the "clean up everything before you exit the function" right 100% of the time.
For code organization, sometimes C++ "concepts" (static polymorphism, so no function pointers) are a nice win. Questionable whether the many lines of code you can save justifies the fact that you're writing template code, however.
The more you look, the more places you'll find clean little wins for C++, with no performance cost.
All change involves risk. You cannot change for the better without some risk. Some ideas are riskier than others, of course, but "actually control immigration for one" seems as conservative as a change can be.
That's the media spin to be sure: anyone against immigration is racist. Also, don't forget the media spin that "Muslim" is a race. It's all bullshit.
Controlling the borders is the duty of any government, regardless of the race of the people on the other side of that border.
Donal Trump is a moderate Democrat on every issue but immigration.
Which only goes to show how deeply, thoroughly corrupt the US has become. We are no longer a nation of laws, but instead have one America for the powerful and celebrities, and another America for the rest of us.
The fact is that a tiny percentage of the population holds the vast majority of the wealth, and they are not spending it.
Wealth is not money; money is not wealth. If you had an even share of all the stock in all the publicly held corporations in the US, it would pay about $1k/year in dividends. Yay? You can't "spend" wealth. You can sell it, at which point you no longer have wealth, and presumably as everyone starts doing that it ends up concentrated in the hands of just a few people again (though it might take a while).
The fact is, the majority of Americans are stockholders, and the median American retires with a reasonable chunk of wealth that she then sells off gradually to live on during that retirement.
Maybe you meant something other than "wealth"? But I can't guess what that would be.
Customer satisfaction is an important goal for any company, but Nintendo especially. High server load for a game during "launch week" requires no ESP - it's entirely predictable. You don't want to build out to a scale that handles launch week load and keep paying for that forever - it can be triple your ongoing load.
If you have an "internal cloud", then you can do this all in your datacenter: provision a ton of capacity, then release it back to your pool for the next game to use. That can be economical. If you don't, then you can borrow capacity from one of the big cloud providers, just pay by the hour until you don't need it.
I do this sort of dynamic provisioning using the cloud for a living these days. It's really not that hard, if your service was built for scale to begin with, and it's quite inexpensive compared to buying servers that will be idle for a year until the next major expansion or whatever.
My current gaming rig is 6 years old. I've switched video cards twice, and I expect a couple more years of use out of it. It's still fast enough, since the vid card does most of the work.
Building your first PC is hard, sure. If you don't enjoy that sort of thing, buy an Alienware. PCs just don't "age out" the way they used to these days: it's no longer something you have to do every 2-3 years.
Are you also going to rent 3x the support staff for a week?
You're not going to have more support issues by provisioning adequate capacity.
Or maybe they just don't like to run their brand new intellectual property on someone else's servers, without an airtight contract, liability for the cloud company, the possibility to audit the provider, etc, etc?
Almost everyone new runs mostly on the cloud these days. Neither Amazon nor Microsoft is going to steal your IP. The whole point of the cloud is to make it trivial to provision large amounts of capacity, temporarily if needed.
Just rent 3x the servers for launch week. There's a whole "cloud" of such servers available. But perhaps innovations from the past 10 years are too new for AAA games.
Can't wait to see kids running around the slums, whore houses, drug dens, and all kinds of shitholes looking for stupid pokemons.
I've been amused by considering the special Pokemon you can only catch in the sewers, or on the grounds of a nuclear power station, or on a military base, or in a burning building, or on a construction site. But sure, crack houses are fun too.
The news is "The problem is that Pokemon Go is not officially available in every region". Shocking that AAA game companies still don't get the single most important fact about selling games. Seriously, WTF?