[xclite]

insertAlias, on 09 March 2011 - 04:53 PM, said:

There's very few times you can ever say "undeniably" to a programmer. People will swear up and down that VIM (or EMACS) is a much better dev tool than any IDE. It's all personal preference, and you can't really say one way or the other which is "better," except in the context of yourself.

That said, the last thing I would do is have new programmers working in notepad and command line compiling. That's just too much information to absorb at once, when you don't have to.

Emacs and VIM are far superior to simple text editors, also. Notepad doesn't offer anything, and it is not a matter of opinion that notepad is inferior to... well pretty much anything for Java. Of course you'll hate it.

Another advantage for C# for new users is the fact that you know what to use - there's no worry about whether or not you are using the best IDE. Install Visual Studio and be done with it.

This post has been edited by xclite: 09 March 2011 - 06:02 PM

[lordofduct]

xclite, on 09 March 2011 - 05:53 PM, said:

Another advantage for C# for new users is the fact that you know what to use - there's no worry about whether or not you are using the best IDE. Install Visual Studio and be done with it.

What about monodevelop...



playing devil's advocate is all

[xclite]

If only it were complete and a true option. One can only dream.

[Patrunjel]

I have seen people complaining about manual memory management in C++.In the other languages there are some tools to do it? Or they just don't let you do it at all? Or they're "dead" after you reach the end of their scope?

[Aphex19]

Ruby was the first language that I used.

print "hello"

[Sergio Tapia]

Patrunjel, on 10 March 2011 - 05:29 AM, said:

I have seen people complaining about manual memory management in C++.In the other languages there are some tools to do it? Or they just don't let you do it at all? Or they're "dead" after you reach the end of their scope?

High level languages have what's called a GarbageCollector.

http://en.wikipedia....uter_science%29

tl;dr: It's smart software that detects when a variable can no longer be used, so it discards it and makes that block of memory available again, **automatically**.

This makes for much nicer to read code.

[xclite]

Sergio Tapia, on 10 March 2011 - 06:40 AM, said:

This makes for much nicer to read code.

At the expense of some performance, but given the errors that can occur when writing your own memory management, the performance cost can be worth it.

This post has been edited by xclite: 10 March 2011 - 07:12 AM

[Sergio Tapia]

According to experienced developers who code in both high and low level languages, most cases, performance is just the same as a low level language like C. Can you give some examples of where manual memory management would be faster than say, something like C# or Java?

[xclite]

Sergio Tapia, on 10 March 2011 - 08:56 AM, said:

According to experienced developers who code in both high and low level languages, most cases, performance is just the same as a low level language like C. Can you give some examples of where manual memory management would be faster than say, something like C# or Java?

Performance in Java can approach and equal the performance of C. However, even without benchmarks, it must by definition cause a performance hit to let the machine determine when and what to collect at runtime rather than just following a free instruction. That hit may be negligible or not even noticeable on modern systems with modern compilers/runtimes, and in fact some styles of garbage collector may be preferable rather than managing memory as you go - if you have natural pauses in execution where you can collect crap, you may want to save it all for the downtime rather than using cycles throughout instruction intensive periods. Different styles of GC have different strengths, clearly.

However, that doesn't prove anything regarding garbage collector performance. That just shows how awesome the optimizations have become. In those examples, it is entirely possible that the GC does come with a hit, but the speeds still approach or are basically the same because of the speedups and optimizations of the JVM, for example.

Manual memory management must be faster, but that speed increase isn't always worth the inconvenience of having to deal with it, nor is it always noticeable or even significant.

This is, of course, way off topic.

[SpeedisaVirus]

In almost all cases its just as good. Most programmers are not as good as they think they are and proper algorithmic approaches to a problem are the performance loss source, not GC (in a mature GC'd language). Anything short of real time performance most programmers couldn't make an appreciable difference in C/C++ compared to Java or C#.

Those are the facts.

[Sergio Tapia]

xclite, on 10 March 2011 - 10:12 AM, said:

Sergio Tapia, on 10 March 2011 - 08:56 AM, said:

According to experienced developers who code in both high and low level languages, most cases, performance is just the same as a low level language like C. Can you give some examples of where manual memory management would be faster than say, something like C# or Java?

Performance in Java can approach and equal the performance of C. However, even without benchmarks, it must by definition cause a performance hit to let the machine determine when and what to collect at runtime rather than just following a free instruction. That hit may be negligible or not even noticeable on modern systems with modern compilers/runtimes, and in fact some styles of garbage collector may be preferable rather than managing memory as you go - if you have natural pauses in execution where you can collect crap, you may want to save it all for the downtime rather than using cycles throughout instruction intensive periods. Different styles of GC have different strengths, clearly.

So you're saying there is a difference, yes. But is the difference noticeable? Will it affect the performance of the application in a noticeable way?

And more importantly: Will that performance offset the business cost and time of the developers?

[Curtis Rutland]

Quote

And more importantly: Will that performance offset the business cost and time of the developers?

I think that depends on what you're writing. A 3d-intensive big name video game? Yeah, it's definitely worth going the more low-level route with C++.

A big business application? Very unlikely that it's worth the trouble.

[xclite]

xclite, on 10 March 2011 - 08:52 AM, said:

Sergio Tapia, on 10 March 2011 - 06:40 AM, said:

This makes for much nicer to read code.

At the expense of some performance, but given the errors that can occur when writing your own memory management, the performance cost can be worth it.

xclite said:

In those examples, it is entirely possible that the GC does come with a hit, but the speeds still approach or are basically the same because of the speedups and optimizations of the JVM, for example.

Manual memory management must be faster, but that speed increase isn't always worth the inconvenience of having to deal with it, nor is it always noticeable or even significant.

SpeedisaVirus, on 10 March 2011 - 09:15 AM, said:

In almost all cases its just as good. Most programmers are not as good as they think they are and proper algorithmic approaches to a problem are the performance loss source, not GC (in a mature GC'd language). Anything short of real time performance most programmers couldn't make an appreciable difference in C/C++ compared to Java or C#.

Those are the facts.

So... basically what I said? Thanks for repeating it.

Sergio Tapia, on 10 March 2011 - 09:19 AM, said:

So you're saying there is a difference, yes. But is the difference noticeable? Will it affect the performance of the application in a noticeable way?

And more importantly: Will that performance offset the business cost and time of the developers?

Depends on the language/GC. Java and C# are mature and it's almost always worth letting them do it for you. Languages where GC is an issue also have other causes of slowness as issues. Try using Ruby for something memory intensive and then watch the garbage collector stop the world.

[Sergio Tapia]

insertAlias, on 10 March 2011 - 10:25 AM, said:

Quote

And more importantly: Will that performance offset the business cost and time of the developers?

I think that depends on what you're writing. A 3d-intensive big name video game? Yeah, it's definitely worth going the more low-level route with C++.

A big business application? Very unlikely that it's worth the trouble.

That's exactly the point I was nudging Xclite towards. And since most of us here aren't working for Blizzard working on the next MMO/Call of Duty...

[xclite]

Sergio Tapia, on 10 March 2011 - 09:41 AM, said:

insertAlias, on 10 March 2011 - 10:25 AM, said:

Quote

And more importantly: Will that performance offset the business cost and time of the developers?

I think that depends on what you're writing. A 3d-intensive big name video game? Yeah, it's definitely worth going the more low-level route with C++.

A big business application? Very unlikely that it's worth the trouble.

That's exactly the point I was nudging Xclite towards. And since most of us here aren't working for Blizzard working on the next MMO/Call of Duty...

I've had it cause issues, but something about classified information doesn't make it susceptible to nudging. Unless you're Bradley Manning

Also I think some people here might like to work for game companies - we've got a pretty big following in the Game Programming forum. Knowing how to manage memory is important, I just wouldn't suggest it first. Which I never did =p

This post has been edited by xclite: 10 March 2011 - 07:52 AM