I've basically given up professional C++ over this. I'd love a smaller project where I get to set the coding standards for a team small enough that I could actually CR everything and keep the rot from setting in, but somehow I've migrated to the big guys where that sort of team doesn't exist.
Dude, you need to be less sensitive if you think I was running down C++ at the expense of Java. Don't make it a religious issue, you won't find happiness along that path.
Sure, I get your point, but there were plenty of fast low-level (or "mid-level" if you want to call it that) languages back in the day, which have mostly faded into obscurity or oblivion now*. C was popular and lasting IMO because it was straightforward, because it felt like a sort of cross-platform macro-assembler. Just a few arcane quirks to figure out for the instructions C didn't represent directly.
*My worst coding project ever was supporting legacy PL/S code. (Made extra fun since IBM never released the compiler so I was bugfixing the generated object.) Ug - give me C any day!
What you are talking about is that we have ABSTRACTED memory management tasks by going through libraries and templates. The "problem" hasn't been "fixed", all we've done is make it easy for programmers to not think about it if they wish.
Yeah, sure, that's all any programming language ever is: layers of abstraction over assembly. My point was, from a practical perspective, you don't worry about memory leaks in modern C++ code any more than you do in modern Java code. But of course Java only offers the "no leaks" way, which in C++ you must discover the "no leaks" way, giving Java the easier learning curve. (You can still get memory leaks with anything, of course, if you try hard enough, but I'm talking about the 99% case here).
But that doesn't mean C++ is old and arcane, nor is it bizarre
Placement new. Weak references. Static initialization quirks (so many lock management and reference counting bugs caused by that over the years!). A Turing-complete templating language, so that you can turn your compiler run into a game of Tetris. Heck, just the fundamental need to understand that vectors will copy their member as they grow, and so you need to use the appropriate kind of smart pointer for that (which has changed, what 3 times now over the life of the language).
Bizarre and arcane. All there for good reason, to solve problems that can't otherwise be solved, but bizarre and arcane.
Java and C# are slower, MUCH slower, and they use a lot more resources when they run.
I used to work on code where that mattered, back in the day. Now everything is I/O bound and horizontally scalable, in my world, and CPU load is around 5% average in my service fleet.
OTOH, guys working in the IOT world with devices with 4K of memory still love C.
Admit it. Java and C# are only tools, just like C and C++
The abstraction in C made writing hand-optimized code a guessing game in the early days. Rather than just writing what you wanted, you had to guess what would be compiled into what you wanted, and check the object code to see if you guessed write. How to rotate a register, how to get both the result and remainder of a single division instruction, that sort of thing. Often people just inlined the desired assembly.
"What you wrote is what you got" was not how veteran assembly devs saw it - there was a lot of arcane C syntax (none of it standardized in the early days) needed to get what you wanted.
None of that is relevant any more, of course. These days the struggle for efficient code is more about understanding the arcane twists of each kernel (e.g., how do I allocate a buffer from userland to ensure no-copy I/O in this environment).
I suppose each person will have to learn this the hard way, like you, before they see the benefit in actually controlling the media.
But that's not a reason to never use plans like this. Just go into it with eyes wide open - you're leasing the content, for an undefined time. If you almost never go back and watch anything more than 5 years old, for example, then this can be a good deal. For movies and music, I personally want control of the media, since I watch old stuff a lot.
For games, OTOH, I'm quite content to buy the game again in 10 years if I need to, since the percentage of old games I go back to is so low. (My favorite games are all 5+ years old, but that's maybe a dozen games, versus the 50 or so I buy each year and never return to). Buying 1% of my games again is totally worth the convenience to me.
For books, I do go back to older books occasionally, but I have ~1000 hardbacks now and I don't have any more freaking room for them. I'm content with the limitations of leasing eBooks, due to simple practical constraints.
Make each decision based on what's best for you, once you understand what you own and what you don't.
Translating to C would not impose a limitation on the language features of C++
Practical limitations, for one guy banging out a language implementation in a hurry. C syntax was kept intact wherever possible, so that no translation would be needed. Which in turn led to quick adoption of C++ by C coders (which doomed C++ code to forever be ruined by C-style coding).
You do realize there are no "memory management issues" in modern C++, right? Java abstracts away the memory management issues of C, and of C++ written like C.
The advantage of Java and C# is the easier learning curve. C++ can be bizarre and arcane, while the managed languages are simpler to get right, and so you can far more easily hire programmers who won't screw everything up.
Mwh, the only thing C gives you over a good macro assembler is parsing of arithmetic expressions, and a somewhat processor-independent syntax. It's a nice layer of abstraction now that compiler optimization is so good, but originally was a real pain if you needed performance. (No one hand hand-code optimal assembly against a modern architecture, so you the performance arguments have swapped ends these days).
Not all Harlem residents are violent either (isn't Harlem gentrified these days?). Obviously, almost all aren't. But it only takes a few to make a violent crime wave.
It's also extremely suspicious timing, right after the UK government demands more powers to snoop on people. While that doesn't disprove anything, the UK government has a history of thing kind of thing
The UK government has a history of placing bombs on airliners? Do you just make up everything now?
The idea was not binary compatibility but source compatibility. Someone in the hierarchy must have dictated that C programs must be able to be recompiled in 64bit with zero code changes. Only an MBA with zero programming background could think that this largely impossible mandate justifies permanently twisting the system with weird rules.
Remember Windows95? The OS that took MS from a bit player to world domination? Yeah, the entire focus of Win95, including the reason it was so unreliable, was this exact sort of compatibility between the 16-bit and 32-bit worlds. Win95 could run 16-bit shared-memory drivers in a 32-bit, protected-memory OS (not safely, but they would work).
Backwards compatibility with 0 code changes is the entire reason anyone today has even heard of MS. Their decline started about the time they abandoned backwards compatibly as a priority.
Oh, you ranted about how you, personally, don't like MS because in your personal experience it was forced on you. Well, I have the same rant about Apple. Different people have different experiences and values, and what a boring world it would be if we were all the same.
The big 5 are Amazon, Apple, Facebook, Google, and Microsoft. All but MS are Apple shops for laptops (and all but MS and Apple are Linux shops otherwise). Amazon and I think Facebook allow MS, but support is second-class and MS is discouraged. Google and obvious Apple only allow MS for specific business needs (competitive analysis, cross-compat testing, etc).
I see the reverse as you, and despise Apple by your same rationale. Of the "big 5" software companies, only at MS itself are you likely to find an environment dominated by MS. I hate having that Apple shit forced on me, itherwise I wouldn't care much about Apple.
I've suggested that if they're common, they might explain the "dark matter" problem of cosmology:namely, a reservoir of matter around galaxies that is impossible to detect by normal means, but doesn't require any exotic, unverified forms of matter to explain.
Sorry, dark matter explains three observations: galactic rotation rates, the balance of matter in the early universe (as observed in the CMBR), and gravitational lensing where there is no visible matter. And it explains the first two of those in a way that matches quantitatively.
We know (as much as we know anything in science) that most of the matter in the early universe was not made from electrons and protons, sorry. Whatever its nature may be, most of the matter in the universe doesn't interact with light in any way.
It also points out the technical flaw in MS's decision. If the "abusers" are filling up that space with multiple backups, just handle that case. Backups are almost never read again. Facebook deals with very-infrequently accessed data by storing it in drives that are powered off. When the data is needed the drive is spun up (access to photos that no one has accessed in quite some time will simply fail to display them, but try again in 5 minutes and they're there). MS could just offer "unlimited, but a long response time if you have more than 1 TB" and likely make people happy.
They advertised unlimited and provided unlimited, now they're warning everyone it's not unlimited, and in a year will stop providing unlimited. There's no way you can twist that to be false advertising.
It's always annoying what a company changes a product in a way you don't like, or raises prices for the same thing, but that has nothing to do with false advertising. Companies that do that excessively are good to avoid, of course, but products do evolve over time.
As far as dealing with spikes in demand, you really have to think that there are two very different environments. There are internal IT departments where demand spikes like that don't exist, and there are public facing operations (ie big web operations) where the cloud makes a lot more sense. I deal with the former, not the latter.
There's a lot of truth to this idea, IMO. Right tools for the job. Moving your Exchange server to the cloud (or Office whatsit where MS does it all) is a very different kind of question than a web service. Moving either your DB or the servers that need it to the cloud, but not both, is consultant-level stupid.
There are other cases too of course. For a software company, using the cloud to do all of your QA makes a lot of sense - I've been at several place that spend years developing cloudlike system to manage pools of servers to allow large QA setups to be shared between teams. A new project like that today internally would be a bit nuts. Another case is a big workload that only happens once a month (e.g., end-of-month accounting runs), which can be a big win to run in the cloud if your auditors allow it.
First, apps must ships files between databases and themselves. In a private datacenter, that's over ethernet. In the cloud, it is often over disparate VPN connection, and worse of all, over your internet connection. How many companies willing to get a 10Gb ethernet pipe?
All my data is next to my servers these days, in the cloud, works well for me. Different tools for different jobs, perhaps?
Now if you are using any shared storage, all bets are off.
I remember the days of fibre channel. Seems like so long ago. What an expensive pain in the ass that was, the bad of days of trying to scale vertically instead of horizontally.
plus cloud provider lock in (just try getting your database out of AWS and see what hidden costs there really are)
True enough, but it's no picnic moving off of Oracle either. It's always going to suck if a key vendor goes bad (but what are the odds anyone will ever go as bad as Oracle?).
Everything's virtual, sure, but hosts fail, have disk problems, eventually age out, etc. It's a pain. I've done it both ways, and I really like having one less category of things to care about. Plus, if you ever need 300 added servers today, and tomorrow, but then the peak will pass, the cloud is nice for that.
Unless they pay for it of course, and then they'll find out running it in house would have been cheaper.
In-house is almost always cheaper if you just look at server costs. But if you move enough stuff to the cloud you can get rid of a large chunk of your IT staff - it's a form of outsourcing. When that's true, you can come out way ahead. This is quite appealing to large companies that do all their IT grunt work through contractors anyhow - it's just a move from one group with contractual SLAs on service to another.
The cloud is also more appealing for services where each tier is a highly-available, load-balanced, stateless sea of servers. That sort of thing really benefits from the trivial and immediate server replacement you have from the cloud providers. (Something's wrong with Server-447? Just drop it and provision a new one, 5 minutes max.) But for simpler services that advantage is lost in the noise of manual software deployment/config/etc to stand up a new box.
This is why I love the Netflix DVD service. It's not timely, but it works out to less than $0.50 for an episode of TV in HD, and I never see any ads (but then, I don't use a consumer DVD player).
You'd be surprised how viewers are tracked these days. Post-broadcast-world viewer data mining is a serious business, and it doesn't only depend on counting streams.
Slashdot Mirror
I've basically given up professional C++ over this. I'd love a smaller project where I get to set the coding standards for a team small enough that I could actually CR everything and keep the rot from setting in, but somehow I've migrated to the big guys where that sort of team doesn't exist.
Dude, you need to be less sensitive if you think I was running down C++ at the expense of Java. Don't make it a religious issue, you won't find happiness along that path.
Sure, I get your point, but there were plenty of fast low-level (or "mid-level" if you want to call it that) languages back in the day, which have mostly faded into obscurity or oblivion now*. C was popular and lasting IMO because it was straightforward, because it felt like a sort of cross-platform macro-assembler. Just a few arcane quirks to figure out for the instructions C didn't represent directly.
*My worst coding project ever was supporting legacy PL/S code. (Made extra fun since IBM never released the compiler so I was bugfixing the generated object.) Ug - give me C any day!
What you are talking about is that we have ABSTRACTED memory management tasks by going through libraries and templates. The "problem" hasn't been "fixed", all we've done is make it easy for programmers to not think about it if they wish.
Yeah, sure, that's all any programming language ever is: layers of abstraction over assembly. My point was, from a practical perspective, you don't worry about memory leaks in modern C++ code any more than you do in modern Java code. But of course Java only offers the "no leaks" way, which in C++ you must discover the "no leaks" way, giving Java the easier learning curve. (You can still get memory leaks with anything, of course, if you try hard enough, but I'm talking about the 99% case here).
But that doesn't mean C++ is old and arcane, nor is it bizarre
Placement new. Weak references. Static initialization quirks (so many lock management and reference counting bugs caused by that over the years!). A Turing-complete templating language, so that you can turn your compiler run into a game of Tetris. Heck, just the fundamental need to understand that vectors will copy their member as they grow, and so you need to use the appropriate kind of smart pointer for that (which has changed, what 3 times now over the life of the language).
Bizarre and arcane. All there for good reason, to solve problems that can't otherwise be solved, but bizarre and arcane.
Java and C# are slower, MUCH slower, and they use a lot more resources when they run.
I used to work on code where that mattered, back in the day. Now everything is I/O bound and horizontally scalable, in my world, and CPU load is around 5% average in my service fleet.
OTOH, guys working in the IOT world with devices with 4K of memory still love C.
Admit it. Java and C# are only tools, just like C and C++
Do you see anyone around here saying different?
The abstraction in C made writing hand-optimized code a guessing game in the early days. Rather than just writing what you wanted, you had to guess what would be compiled into what you wanted, and check the object code to see if you guessed write. How to rotate a register, how to get both the result and remainder of a single division instruction, that sort of thing. Often people just inlined the desired assembly.
"What you wrote is what you got" was not how veteran assembly devs saw it - there was a lot of arcane C syntax (none of it standardized in the early days) needed to get what you wanted.
None of that is relevant any more, of course. These days the struggle for efficient code is more about understanding the arcane twists of each kernel (e.g., how do I allocate a buffer from userland to ensure no-copy I/O in this environment).
I suppose each person will have to learn this the hard way, like you, before they see the benefit in actually controlling the media.
But that's not a reason to never use plans like this. Just go into it with eyes wide open - you're leasing the content, for an undefined time. If you almost never go back and watch anything more than 5 years old, for example, then this can be a good deal. For movies and music, I personally want control of the media, since I watch old stuff a lot.
For games, OTOH, I'm quite content to buy the game again in 10 years if I need to, since the percentage of old games I go back to is so low. (My favorite games are all 5+ years old, but that's maybe a dozen games, versus the 50 or so I buy each year and never return to). Buying 1% of my games again is totally worth the convenience to me.
For books, I do go back to older books occasionally, but I have ~1000 hardbacks now and I don't have any more freaking room for them. I'm content with the limitations of leasing eBooks, due to simple practical constraints.
Make each decision based on what's best for you, once you understand what you own and what you don't.
Translating to C would not impose a limitation on the language features of C++
Practical limitations, for one guy banging out a language implementation in a hurry. C syntax was kept intact wherever possible, so that no translation would be needed. Which in turn led to quick adoption of C++ by C coders (which doomed C++ code to forever be ruined by C-style coding).
You do realize there are no "memory management issues" in modern C++, right? Java abstracts away the memory management issues of C, and of C++ written like C.
The advantage of Java and C# is the easier learning curve. C++ can be bizarre and arcane, while the managed languages are simpler to get right, and so you can far more easily hire programmers who won't screw everything up.
Mwh, the only thing C gives you over a good macro assembler is parsing of arithmetic expressions, and a somewhat processor-independent syntax. It's a nice layer of abstraction now that compiler optimization is so good, but originally was a real pain if you needed performance. (No one hand hand-code optimal assembly against a modern architecture, so you the performance arguments have swapped ends these days).
Not all Harlem residents are violent either (isn't Harlem gentrified these days?). Obviously, almost all aren't. But it only takes a few to make a violent crime wave.
It's also extremely suspicious timing, right after the UK government demands more powers to snoop on people. While that doesn't disprove anything, the UK government has a history of thing kind of thing
The UK government has a history of placing bombs on airliners? Do you just make up everything now?
The idea was not binary compatibility but source compatibility. Someone in the hierarchy must have dictated that C programs must be able to be recompiled in 64bit with zero code changes. Only an MBA with zero programming background could think that this largely impossible mandate justifies permanently twisting the system with weird rules.
Remember Windows95? The OS that took MS from a bit player to world domination? Yeah, the entire focus of Win95, including the reason it was so unreliable, was this exact sort of compatibility between the 16-bit and 32-bit worlds. Win95 could run 16-bit shared-memory drivers in a 32-bit, protected-memory OS (not safely, but they would work).
Backwards compatibility with 0 code changes is the entire reason anyone today has even heard of MS. Their decline started about the time they abandoned backwards compatibly as a priority.
Oh, you ranted about how you, personally, don't like MS because in your personal experience it was forced on you. Well, I have the same rant about Apple. Different people have different experiences and values, and what a boring world it would be if we were all the same.
I speak only to my personal experience, not the rest of America.
The big 5 are Amazon, Apple, Facebook, Google, and Microsoft. All but MS are Apple shops for laptops (and all but MS and Apple are Linux shops otherwise). Amazon and I think Facebook allow MS, but support is second-class and MS is discouraged. Google and obvious Apple only allow MS for specific business needs (competitive analysis, cross-compat testing, etc).
I see the reverse as you, and despise Apple by your same rationale. Of the "big 5" software companies, only at MS itself are you likely to find an environment dominated by MS. I hate having that Apple shit forced on me, itherwise I wouldn't care much about Apple.
I've suggested that if they're common, they might explain the "dark matter" problem of cosmology:namely, a reservoir of matter around galaxies that is impossible to detect by normal means, but doesn't require any exotic, unverified forms of matter to explain.
Sorry, dark matter explains three observations: galactic rotation rates, the balance of matter in the early universe (as observed in the CMBR), and gravitational lensing where there is no visible matter. And it explains the first two of those in a way that matches quantitatively.
We know (as much as we know anything in science) that most of the matter in the early universe was not made from electrons and protons, sorry. Whatever its nature may be, most of the matter in the universe doesn't interact with light in any way.
Mod up AC - very informative.
It also points out the technical flaw in MS's decision. If the "abusers" are filling up that space with multiple backups, just handle that case. Backups are almost never read again. Facebook deals with very-infrequently accessed data by storing it in drives that are powered off. When the data is needed the drive is spun up (access to photos that no one has accessed in quite some time will simply fail to display them, but try again in 5 minutes and they're there). MS could just offer "unlimited, but a long response time if you have more than 1 TB" and likely make people happy.
They advertised unlimited and provided unlimited, now they're warning everyone it's not unlimited, and in a year will stop providing unlimited. There's no way you can twist that to be false advertising.
It's always annoying what a company changes a product in a way you don't like, or raises prices for the same thing, but that has nothing to do with false advertising. Companies that do that excessively are good to avoid, of course, but products do evolve over time.
As far as dealing with spikes in demand, you really have to think that there are two very different environments. There are internal IT departments where demand spikes like that don't exist, and there are public facing operations (ie big web operations) where the cloud makes a lot more sense. I deal with the former, not the latter.
There's a lot of truth to this idea, IMO. Right tools for the job. Moving your Exchange server to the cloud (or Office whatsit where MS does it all) is a very different kind of question than a web service. Moving either your DB or the servers that need it to the cloud, but not both, is consultant-level stupid.
There are other cases too of course. For a software company, using the cloud to do all of your QA makes a lot of sense - I've been at several place that spend years developing cloudlike system to manage pools of servers to allow large QA setups to be shared between teams. A new project like that today internally would be a bit nuts. Another case is a big workload that only happens once a month (e.g., end-of-month accounting runs), which can be a big win to run in the cloud if your auditors allow it.
First, apps must ships files between databases and themselves. In a private datacenter, that's over ethernet. In the cloud, it is often over disparate VPN connection, and worse of all, over your internet connection. How many companies willing to get a 10Gb ethernet pipe?
All my data is next to my servers these days, in the cloud, works well for me. Different tools for different jobs, perhaps?
Now if you are using any shared storage, all bets are off.
I remember the days of fibre channel. Seems like so long ago. What an expensive pain in the ass that was, the bad of days of trying to scale vertically instead of horizontally.
plus cloud provider lock in (just try getting your database out of AWS and see what hidden costs there really are)
True enough, but it's no picnic moving off of Oracle either. It's always going to suck if a key vendor goes bad (but what are the odds anyone will ever go as bad as Oracle?).
Everything's virtual, sure, but hosts fail, have disk problems, eventually age out, etc. It's a pain. I've done it both ways, and I really like having one less category of things to care about. Plus, if you ever need 300 added servers today, and tomorrow, but then the peak will pass, the cloud is nice for that.
Unless they pay for it of course, and then they'll find out running it in house would have been cheaper.
In-house is almost always cheaper if you just look at server costs. But if you move enough stuff to the cloud you can get rid of a large chunk of your IT staff - it's a form of outsourcing. When that's true, you can come out way ahead. This is quite appealing to large companies that do all their IT grunt work through contractors anyhow - it's just a move from one group with contractual SLAs on service to another.
The cloud is also more appealing for services where each tier is a highly-available, load-balanced, stateless sea of servers. That sort of thing really benefits from the trivial and immediate server replacement you have from the cloud providers. (Something's wrong with Server-447? Just drop it and provision a new one, 5 minutes max.) But for simpler services that advantage is lost in the noise of manual software deployment/config/etc to stand up a new box.
This is why I love the Netflix DVD service. It's not timely, but it works out to less than $0.50 for an episode of TV in HD, and I never see any ads (but then, I don't use a consumer DVD player).
You'd be surprised how viewers are tracked these days. Post-broadcast-world viewer data mining is a serious business, and it doesn't only depend on counting streams.