|
Thread Rules
1. This is not a "do my homework for me" thread. If you have specific questions, ask, but don't post an assignment or homework problem and expect an exact solution.
2. No recruiting for your cockamamie projects (you won't replace facebook with 3 dudes you found on the internet and $20)
3. If you can't articulate why a language is bad, don't start slinging shit about it. Just remember that nothing is worse than making CSS IE6 compatible.
4. Use [code] tags to format code blocks. |
On November 14 2012 08:28 meatpudding wrote:Show nested quote +On November 14 2012 07:52 KaiserJohan wrote:On November 13 2012 08:37 Nausea wrote:
Okay so I started writing this assetmanager singleton class, with a map storing a string and an image. The problem is that the image is stored as a pointer in the map and adding it to the map is going fine, however when i try to get the function to return the pointer the compiler does not complain but the program shuts down when the pointer is returned as NULL.
Help! :/
NVM I solved it, I tried doing it->second, and instead I now did ImagesList.find(filename)->second
that did the trick. Thank you all for your help Now I'm mainly speaking from a C++ perspective but: Drop the singleton design, instead build your system in such a way as to avoid globals as much as absolutely possible. It forces you to think and create healthy software design, because sooner or later the singleton and stuff like it will bite you. Your object should have a clear purpose and their scope and visibility as limited as possible and also avoid coupling between systems. I would like to ask how you suggest doing this? Imagine different Entity types need access to the AssetManager. With globals or singleton you can do something like this (Assume myImage is inherit from Entity and myImageResourceName identifies the unique resource)
// For the projectile class
void Projectile::Load()
{
myImage = AssetManager::GetSingleton().GetResource(myImageResourceName);
}
// For the shrubbery class
void Shrubbery::Load()
{
myImage = AssetManager::GetSingleton().GetResource(myImageResourceName);
}
How would you avoid AssetManager being global in this case?
Don't know which language, so pseudocode:
Projectile::constructor(AssetManager am) { this,assetManager = am; }
Projectile::Load() { this,assetManager.GetResource(imageName); }
|
I love hearing people blindly saying you should not use globals. Its one of those things that gets taught in schools and blindly accepted. Globals are fine when used correctly, people are taught not to use them because they present some risk to maintainability especially when they are used in the wrong context.
As for your loading problem, this is a poor case to do lazy loading. If you are running your game loop and you need a resource for the first time you do not want to incur the cost of going to disk to load it because it would make your frame drop and make the controls feel inconstant/lagy. This is why games have loading times at the start of a level, they are pulling all the resources into memory ahead of time so they do not have to take that hit in the middle of game play.
That all being said you also have to consider the total memory foot print of your program. This is a very complicated topic, but overly simplified: you want it as small as possible. For this reason only load what you need into memory at the start of a level, and unload it when it is no longer needed.
|
On November 14 2012 07:52 KaiserJohan wrote:Show nested quote +On November 13 2012 08:37 Nausea wrote:
Okay so I started writing this assetmanager singleton class, with a map storing a string and an image. The problem is that the image is stored as a pointer in the map and adding it to the map is going fine, however when i try to get the function to return the pointer the compiler does not complain but the program shuts down when the pointer is returned as NULL.
Help! :/
NVM I solved it, I tried doing it->second, and instead I now did ImagesList.find(filename)->second
that did the trick. Thank you all for your help Now I'm mainly speaking from a C++ perspective but: Drop the singleton design, instead build your system in such a way as to avoid globals as much as absolutely possible. It forces you to think and create healthy software design, because sooner or later the singleton and stuff like it will bite you. Your object should have a clear purpose and their scope and visibility as limited as possible and also avoid coupling between systems.
Singletons are perfectly fine, you just need to legitimately understand how to use them. If anything, for a side project, it's probably better he experiences the first hand the limitations and requirements of using the singleton pattern.
Anyways, to post a general question:
I am looking to start learning Objective C for iPad development, but I am not thrilled about Apple's tutorials for this (I'd really rather not spend 8 hours dragging and dropping things), does anyone have a suggestion for a good place to start learning?
|
On November 12 2012 14:24 ninjamyst wrote:
I am currently a Software Architect at a mid-size Healthcare IT firm. Straight out of college, we pay $55-$60k. 2-3 years of experience will get you $75-$85k. 4-6 years of experience, you are looking at $90-$110k. That doesn't include year-end bonus of $5-$20k. This is in a major city, and we primarily use Microsoft technology....so C#, ASP.NET, SQL Server 2008, etc. We don't care if you memorized definitions, and we dont give you stupid brain teasers during our interviews. We hire based on your problem solving skills and your object oriented design experiences. We will ask you to take a simple every day concept and break it down into classes/methods/properties. For example, how would you represent a baseball game in code? We dont hire specifically for front end vs database vs business logic. As part of our small development team, you will do all of those things. You will learn how to create GUI, how to create business objects, how to maintain databases.
For those of you in college now, the best way to impress is to be involved in side projects outside of school. Most schools don't even teach C# or PHP or Ruby; they only teach Java or C++. When we look at college resumes, everyone takes the same classes, do the same senior projects, and etc. We look for people that wrote their own mobile app, engaged in engineering competitions, did side projects for fun, learned a language outside of class, and etc.
Most importantly, we look for people with strong communication skills. Too often, we get people obsessed with technical stuff only, and they lack social skills. In a team environment, you need to be able to express your ideas to your peers. In design sessions, you need to be able to clearly explain your logic and your reasoning. If we can't understand you, we can't hire you even if you can create the fastest algorithm.
That sounds amazeballs.
I did make an iphone game during my time at Uni and do some extra cirricular volunteer work though. You think that warrants a $50k Salary?
|
|
|
On November 14 2012 09:21 NukeTheBunnys wrote:
I love hearing people blindly saying you should not use globals. Its one of those things that gets taught in schools and blindly accepted. Globals are fine when used correctly, people are taught not to use them because they present some risk to maintainability especially when they are used in the wrong context.
Its not so much maintainability as it is testability where globals cause issues. It is much better for testing to explicitly specify what your class' dependencies are and pass them in, than for a class to assume that a global variable of a specific type is set and initialized properly and utilize that (and in many cases having a global dependency makes it hard/impossible to replace with a stubbed version of the dependency). It also makes your code flow and initialization of objects much more straight-forward and easy to follow. Yes, they're fine when used in the right situations, but there are very few situations where that's the case.
|
On November 14 2012 13:09 tec27 wrote:Show nested quote +On November 14 2012 09:21 NukeTheBunnys wrote:
I love hearing people blindly saying you should not use globals. Its one of those things that gets taught in schools and blindly accepted. Globals are fine when used correctly, people are taught not to use them because they present some risk to maintainability especially when they are used in the wrong context. Its not so much maintainability as it is testability where globals cause issues. It is much better for testing to explicitly specify what your class' dependencies are and pass them in, than for a class to assume that a global variable of a specific type is set and initialized properly and utilize that (and in many cases having a global dependency makes it hard/impossible to replace with a stubbed version of the dependency). It also makes your code flow and initialization of objects much more straight-forward and easy to follow. Yes, they're fine when used in the right situations, but there are very few situations where that's the case.
I agree, Globals make unit testing trickier. Avoid if you can, but use them if you need a quickfix.
|
I was asked this question for an interview and I've been wondering about it for a few weeks.
Rank each device from highest electricity consumption to lowest.
Normal Hard Drive, SSD, USB Flash Drive, RAM
My order of ranking and reasoning:
1) Hard Drive - has to spin at thousands of RPM
2) SSD - Bigger than the other two smaller components of memory [RAM and USB]
3) USB Flash Drive - A usb port can supply up to 5v iirc?
4) Ram - Takes like... 1.5v?
I'm not sure about the last two. Does a USB drive take more electricity or RAM?
|
On November 14 2012 08:34 Morfildur wrote:Show nested quote +On November 14 2012 08:28 meatpudding wrote:On November 14 2012 07:52 KaiserJohan wrote:On November 13 2012 08:37 Nausea wrote:
Okay so I started writing this assetmanager singleton class, with a map storing a string and an image. The problem is that the image is stored as a pointer in the map and adding it to the map is going fine, however when i try to get the function to return the pointer the compiler does not complain but the program shuts down when the pointer is returned as NULL.
Help! :/
NVM I solved it, I tried doing it->second, and instead I now did ImagesList.find(filename)->second
that did the trick. Thank you all for your help Now I'm mainly speaking from a C++ perspective but: Drop the singleton design, instead build your system in such a way as to avoid globals as much as absolutely possible. It forces you to think and create healthy software design, because sooner or later the singleton and stuff like it will bite you. Your object should have a clear purpose and their scope and visibility as limited as possible and also avoid coupling between systems. I would like to ask how you suggest doing this? Imagine different Entity types need access to the AssetManager. With globals or singleton you can do something like this (Assume myImage is inherit from Entity and myImageResourceName identifies the unique resource)
// For the projectile class
void Projectile::Load()
{
myImage = AssetManager::GetSingleton().GetResource(myImageResourceName);
}
// For the shrubbery class
void Shrubbery::Load()
{
myImage = AssetManager::GetSingleton().GetResource(myImageResourceName);
}
How would you avoid AssetManager being global in this case? Don't know which language, so pseudocode:
Projectile::constructor(AssetManager am) { this,assetManager = am; }
Projectile::Load() { this,assetManager.GetResource(imageName); }
Is this not almost exactly how a singleton works?
You create an instance of it once, assign the this pointer and from then on whenever it constructs it just uses the pointer? Am i missing something here?
|
On November 14 2012 13:57 chaokel wrote:Show nested quote +On November 14 2012 08:34 Morfildur wrote:On November 14 2012 08:28 meatpudding wrote:On November 14 2012 07:52 KaiserJohan wrote:On November 13 2012 08:37 Nausea wrote:
Okay so I started writing this assetmanager singleton class, with a map storing a string and an image. The problem is that the image is stored as a pointer in the map and adding it to the map is going fine, however when i try to get the function to return the pointer the compiler does not complain but the program shuts down when the pointer is returned as NULL.
Help! :/
NVM I solved it, I tried doing it->second, and instead I now did ImagesList.find(filename)->second
that did the trick. Thank you all for your help Now I'm mainly speaking from a C++ perspective but: Drop the singleton design, instead build your system in such a way as to avoid globals as much as absolutely possible. It forces you to think and create healthy software design, because sooner or later the singleton and stuff like it will bite you. Your object should have a clear purpose and their scope and visibility as limited as possible and also avoid coupling between systems. I would like to ask how you suggest doing this? Imagine different Entity types need access to the AssetManager. With globals or singleton you can do something like this (Assume myImage is inherit from Entity and myImageResourceName identifies the unique resource)
// For the projectile class
void Projectile::Load()
{
myImage = AssetManager::GetSingleton().GetResource(myImageResourceName);
}
// For the shrubbery class
void Shrubbery::Load()
{
myImage = AssetManager::GetSingleton().GetResource(myImageResourceName);
}
How would you avoid AssetManager being global in this case? Don't know which language, so pseudocode:
Projectile::constructor(AssetManager am) { this,assetManager = am; }
Projectile::Load() { this,assetManager.GetResource(imageName); }
Is this not almost exactly how a singleton works? You create an instance of it once, assign the this pointer and from then on whenever it constructs it just uses the pointer? Am i missing something here?
The end result is the same in this case, but the benefit to this approach is that its far more testable. If you need to test the Projectile class you can create whatever sort of AssetManager suits your testcase (a mock, one that loads things from memory instead of disk, etc.). In the actual game, you can utilize a different one. If you use a singleton approach, you've coupled your Projectile class to whatever AssetManager type it creates, and this obviously makes unit testing harder and makes your underlying types harder to switch around down the road. In general you should prefer to pass dependencies in rather than creating them within each dependent class (although this applies to everything, not just singletons).
|
Ohhh that makes a lot of sense, thanks!
|
On November 14 2012 13:26 frogmelter wrote:
I was asked this question for an interview and I've been wondering about it for a few weeks.
Rank each device from highest electricity consumption to lowest.
Normal Hard Drive, SSD, USB Flash Drive, RAM
My order of ranking and reasoning:
1) Hard Drive - has to spin at thousands of RPM
2) SSD - Bigger than the other two smaller components of memory [RAM and USB]
3) USB Flash Drive - A usb port can supply up to 5v iirc?
4) Ram - Takes like... 1.5v?
I'm not sure about the last two. Does a USB drive take more electricity or RAM?
voltage isn't a great indicator of power consumption. power = voltage * current, so your only getting half of the equation. Also it really depends whether your talking about average power during normal use or while stressing the device with reads/writes.
The actual numbers are something like:
RAM ~ 10W per stick
HDD ~ 10W
SDD ~ 1W
USB <1W
|
On November 14 2012 08:28 meatpudding wrote:Show nested quote +On November 14 2012 07:52 KaiserJohan wrote:On November 13 2012 08:37 Nausea wrote:
Okay so I started writing this assetmanager singleton class, with a map storing a string and an image. The problem is that the image is stored as a pointer in the map and adding it to the map is going fine, however when i try to get the function to return the pointer the compiler does not complain but the program shuts down when the pointer is returned as NULL.
Help! :/
NVM I solved it, I tried doing it->second, and instead I now did ImagesList.find(filename)->second
that did the trick. Thank you all for your help Now I'm mainly speaking from a C++ perspective but: Drop the singleton design, instead build your system in such a way as to avoid globals as much as absolutely possible. It forces you to think and create healthy software design, because sooner or later the singleton and stuff like it will bite you. Your object should have a clear purpose and their scope and visibility as limited as possible and also avoid coupling between systems. I would like to ask how you suggest doing this? Imagine different Entity types need access to the AssetManager. With globals or singleton you can do something like this (Assume myImage is inherit from Entity and myImageResourceName identifies the unique resource)
// For the projectile class
void Projectile::Load()
{
myImage = AssetManager::GetSingleton().GetResource(myImageResourceName);
}
// For the shrubbery class
void Shrubbery::Load()
{
myImage = AssetManager::GetSingleton().GetResource(myImageResourceName);
}
How would you avoid AssetManager being global in this case?
of the top of my mind you could use a template for assetmanager.
|
On November 14 2012 13:57 chaokel wrote:Show nested quote +On November 14 2012 08:34 Morfildur wrote:On November 14 2012 08:28 meatpudding wrote:On November 14 2012 07:52 KaiserJohan wrote:On November 13 2012 08:37 Nausea wrote:
Okay so I started writing this assetmanager singleton class, with a map storing a string and an image. The problem is that the image is stored as a pointer in the map and adding it to the map is going fine, however when i try to get the function to return the pointer the compiler does not complain but the program shuts down when the pointer is returned as NULL.
Help! :/
NVM I solved it, I tried doing it->second, and instead I now did ImagesList.find(filename)->second
that did the trick. Thank you all for your help Now I'm mainly speaking from a C++ perspective but: Drop the singleton design, instead build your system in such a way as to avoid globals as much as absolutely possible. It forces you to think and create healthy software design, because sooner or later the singleton and stuff like it will bite you. Your object should have a clear purpose and their scope and visibility as limited as possible and also avoid coupling between systems. I would like to ask how you suggest doing this? Imagine different Entity types need access to the AssetManager. With globals or singleton you can do something like this (Assume myImage is inherit from Entity and myImageResourceName identifies the unique resource)
// For the projectile class
void Projectile::Load()
{
myImage = AssetManager::GetSingleton().GetResource(myImageResourceName);
}
// For the shrubbery class
void Shrubbery::Load()
{
myImage = AssetManager::GetSingleton().GetResource(myImageResourceName);
}
How would you avoid AssetManager being global in this case? Don't know which language, so pseudocode:
Projectile::constructor(AssetManager am) { this,assetManager = am; }
Projectile::Load() { this,assetManager.GetResource(imageName); }
Is this not almost exactly how a singleton works? You create an instance of it once, assign the this pointer and from then on whenever it constructs it just uses the pointer? Am i missing something here?
No. It's called dependency injection.
http://en.wikipedia.org/wiki/Dependency_Injection
For example if you want to write a unit test, you can pass a derived class of the AssetManager - e.g. a TestAssetManager - and then your class will use that instead. Your program can also use different asset managers that all inherit from the AssetManager class without your Projectile class having to know about that.
|
Interesting discussion, makes one learn something new 
Maybe if he wrote something like this:
or does this design implies into some kind of difficulty?
|
On November 14 2012 21:47 fabiano wrote:Interesting discussion, makes one learn something new Maybe if he wrote something like this: or does this design implies into some kind of difficulty?
It's basically the same, except it uses an explicit interface for the AssetManager. C++ has no notion of Interfaces so it boils down to the same code.
|
On November 14 2012 13:09 tec27 wrote:Show nested quote +On November 14 2012 09:21 NukeTheBunnys wrote:
I love hearing people blindly saying you should not use globals. Its one of those things that gets taught in schools and blindly accepted. Globals are fine when used correctly, people are taught not to use them because they present some risk to maintainability especially when they are used in the wrong context. Its not so much maintainability as it is testability where globals cause issues. It is much better for testing to explicitly specify what your class' dependencies are and pass them in, than for a class to assume that a global variable of a specific type is set and initialized properly and utilize that (and in many cases having a global dependency makes it hard/impossible to replace with a stubbed version of the dependency). It also makes your code flow and initialization of objects much more straight-forward and easy to follow. Yes, they're fine when used in the right situations, but there are very few situations where that's the case.
The real problem with globals and singletons (which are used like globals, as opposed to their namesake, which can be handled by factories just as well) is that they solve a problem that shouldn't exist in the first place.
All that globals solve is lazyness (too lazy to pass around one extra parameter in functions that need access to the global), while providing no additional advantages, outside of a few ridiculously limited side cases (e.g. programs where the additional push to the stack in a function call for the extra parameter actually has a measurable effect on performance, which is almost never the case).
On the other hand, they introduce a lot of downsides in the long term. There is not a production code in the world that has been handled by more than one or two people over a timespan of more than a year, where globals or singletons haven't caused some sort of problem (even if it's miniscule) or made some refactoring necessary down the line. They fuck up unit tests, introduce problems with parallelism (especially pure globals), introduce opaque dependencies, etc.
And again, literally all they solve is lazyness. There's nothing that a global or singleton can do, that you can't do by passing stuff around in parameters.
The reason that universities and schools blindly teach you "don't use globals" is because it's true: You shouldn't. If you're lucky, the global/singleton will be equally as good as the solution without. But it's never better, and often times worse. So just don't use them.
|
Instead of Dependency Injection by using constructor injection, you could also use the Service Locator pattern. They kind-of achieve the same thing: decoupling your classes from concrete implementations.
A C# example:
// somewhere during program startup
ServiceLocator.Register<IAssetManager>(new AssetManager());
// Consumer
IAssetManager am = ServiceLocator.Resolve<IAssetManager>();
am.GetResource(myImageResourceName);
|
1019 Posts
On November 12 2012 14:24 ninjamyst wrote:
I am currently a Software Architect at a mid-size Healthcare IT firm. Straight out of college, we pay $55-$60k. 2-3 years of experience will get you $75-$85k. 4-6 years of experience, you are looking at $90-$110k. That doesn't include year-end bonus of $5-$20k. This is in a major city, and we primarily use Microsoft technology....so C#, ASP.NET, SQL Server 2008, etc. We don't care if you memorized definitions, and we dont give you stupid brain teasers during our interviews. We hire based on your problem solving skills and your object oriented design experiences. We will ask you to take a simple every day concept and break it down into classes/methods/properties. For example, how would you represent a baseball game in code? We dont hire specifically for front end vs database vs business logic. As part of our small development team, you will do all of those things. You will learn how to create GUI, how to create business objects, how to maintain databases.
For those of you in college now, the best way to impress is to be involved in side projects outside of school. Most schools don't even teach C# or PHP or Ruby; they only teach Java or C++. When we look at college resumes, everyone takes the same classes, do the same senior projects, and etc. We look for people that wrote their own mobile app, engaged in engineering competitions, did side projects for fun, learned a language outside of class, and etc.
Most importantly, we look for people with strong communication skills. Too often, we get people obsessed with technical stuff only, and they lack social skills. In a team environment, you need to be able to express your ideas to your peers. In design sessions, you need to be able to clearly explain your logic and your reasoning. If we can't understand you, we can't hire you even if you can create the fastest algorithm.
How hard is it to learn SQL, PHP, C# after you learn C++? I'm an intermediate c++ programmer. It looks like it would be easier since you know a little bit about programming already.
|
On November 15 2012 02:43 white_horse wrote:Show nested quote +On November 12 2012 14:24 ninjamyst wrote:
I am currently a Software Architect at a mid-size Healthcare IT firm. Straight out of college, we pay $55-$60k. 2-3 years of experience will get you $75-$85k. 4-6 years of experience, you are looking at $90-$110k. That doesn't include year-end bonus of $5-$20k. This is in a major city, and we primarily use Microsoft technology....so C#, ASP.NET, SQL Server 2008, etc. We don't care if you memorized definitions, and we dont give you stupid brain teasers during our interviews. We hire based on your problem solving skills and your object oriented design experiences. We will ask you to take a simple every day concept and break it down into classes/methods/properties. For example, how would you represent a baseball game in code? We dont hire specifically for front end vs database vs business logic. As part of our small development team, you will do all of those things. You will learn how to create GUI, how to create business objects, how to maintain databases.
For those of you in college now, the best way to impress is to be involved in side projects outside of school. Most schools don't even teach C# or PHP or Ruby; they only teach Java or C++. When we look at college resumes, everyone takes the same classes, do the same senior projects, and etc. We look for people that wrote their own mobile app, engaged in engineering competitions, did side projects for fun, learned a language outside of class, and etc.
Most importantly, we look for people with strong communication skills. Too often, we get people obsessed with technical stuff only, and they lack social skills. In a team environment, you need to be able to express your ideas to your peers. In design sessions, you need to be able to clearly explain your logic and your reasoning. If we can't understand you, we can't hire you even if you can create the fastest algorithm. How hard is it to learn SQL, PHP, C# after you learn C++? I'm an intermediate c++ programmer. It looks like it would be easier since you know a little bit about programming already.
It's not hard, but every language has small details that don't exist in other languages. For example, PHP has a foreach loop that does not exist in C++ ( it might in the latest version of C++... not sure).
The only languages that are going to be difficult after C++ are:
- Assembly
- C (Because C++ programmers don't know anything about C, but think they do)
- Functional Programming languages like Scheme.
- HDL languages like Verilog and VHDL ( but these can be irrelevant)
|
|
|
|
|
|
EPT![[image loading]](http://i.imgur.com/jGVUT.png)
