Does C++ finally have enough features to make memory leaks a ... memory?
-
Granted I haven't done much C++ (even though it was a long while ago I did tons more C). But I am under the impression that the memory problem in C++ and C are related to the same problem... It is just that C program often do simpler things than the more complicated C++ one. To be more specific the most difficult numerical algorithm might very well be implemented in C, but the program with the most object or bits of memory moving around are likely to be in C++..... So it's premature to boast C superiority!
A new .NET Serializer All in one Menu-Ribbon Bar Taking over the world since 1371!
Super Lloyd wrote:
But I am under the impression that the memory problem in C++ and C are related to the same problem...
That is correct. Both allow the programmer to control the scope in which an allocation lives. So if the programmer fails to control that then an actual leak is produced. More so both also allow programmers explicit write control over memory. That makes more serious problems possible. Those are the ones that generally lead to security problems where as memory leaks (in the strict definition) most often just lead to business functionality problems. Other languages like C# and Java can both have memory leaks (strict definition) but is less likely to occur.
-
With C# and .Net I have seen way too many cases of idiotic memory leaks, just because the developers had no concept of an object lifecycle at all. I have also seen an entire team of Java guys sitting around a server that needed a reset every day because their opus hogged memory. If anything at all, the Java Hobbits are even more religious about not managing your memory. And what could all those senior devs, architects and project leads do about the the memory leak? They beat the garbage collection into submission and fixed the symptom - for a while. Garbage collection and all sorts of automated memory management can't replace a proper architecture and object lifecycles. At best it makes routine work a little easier and offers at best a safety net. At the same time it makes the developers ignorant and careless. The Java guys have made a religion out of it. As long as I use C++, I simply have put debug assertions into the code that monitor what is going on on the heap and fire when there seems to be a leak. After that it's only a little detective work to identify the type of objects that are piling up and then take a good look at the lifecycle of these objects. It's actually not so hard and you don't need any fancy features. The only thing you need is to invest a little more thought and give up your 'right' to write careless code. Then again, a little more thinking, discipline and a little less cowboy coding would also solve many other problems.
I have lived with several Zen masters - all of them were cats. His last invention was an evil Lasagna. It didn't kill anyone, and it actually tasted pretty good.
CodeWraith wrote:
I have also seen an entire team of Java guys sitting around a server that needed a reset every day because their opus hogged memory.
I was told by a number of unix admins that when C/C++ servers were the norm that was the standard operating procedure for any large enterprise. It was something that was planned for when putting the process together for a new server application. And it wasn't just for memory. File descriptors and sockets also needed to be cleaned up.
-
When I read you message I have a feeling that you have no idea what is the causes of memory leak in C#.... For one thing it has little to do with "object lifecyle" FYI, unless you do interop, which is less and less common those days, the only reason for leak in C# is if your referenced by static variable or (though is the same thing in disguise) static event. Or maybe too many live threads which don't die (and all the memory they collect), could typically come from wait handle hanging forever...
A new .NET Serializer All in one Menu-Ribbon Bar Taking over the world since 1371!
Super Lloyd wrote:
For one thing it has little to do with "object lifecyle"
Perhaps we have a different definition of that term. I have seen problems in both Java and C# where the developers failed to properly account for how an object 'ended'. They had no problem with when the object 'started' but they never considered how to specifically control how it 'ended'. Matter of fact I suspect that I have seen code that might have that sort of problem just in the past week.
Super Lloyd wrote:
FYI, unless you do interop, which is less and less common those days, the only reason for leak in C# is if your referenced by static variable or (though is the same thing in disguise) static event. Or maybe too many live threads which don't die (and all the memory they collect), could typically come from wait handle hanging forever...
Probably been at least 15 years since I worked in a system where at least some threads did not die. Often quite a few of them. At least in Java the server frameworks build that in extensively. Most problems I have seen have been because the developers did not even seem to be aware that they were working in a threaded environment. Java and C# both have complex thread pool support and the concept has existed since before C# existed. Might have existed before Java but I wasn't introduced to it until after I first used Java. For context I started with Java 1.1.4 (well before 1.4.)
-
CodeWraith wrote:
I have also seen an entire team of Java guys sitting around a server that needed a reset every day because their opus hogged memory.
I was told by a number of unix admins that when C/C++ servers were the norm that was the standard operating procedure for any large enterprise. It was something that was planned for when putting the process together for a new server application. And it wasn't just for memory. File descriptors and sockets also needed to be cleaned up.
I think to remember for the first windows servers that this was a standard procedure (not daily, but about weekely). This not because of the applications/services but because windows was the problem at that time.
It does not solve my Problem, but it answers my question
-
:-D I should say that I had "early retired" back when auto_ptr was the new k3wl thing. I understand that after that, there have been some other features added that addressed issues that auto_ptr did not completely fix. I wonder if the state of the language is such that there are no more issues with memory leaks - which would indicate to me that no one need bother with calling delete anymore, basically making C++ like its managed cousin C#. Or am I missing something?
To be honest, I like the "feature" that I as the programmer is responsible for each and every resource and therefore need to pay attention to it. This is a clear straight forward situation and in case I'm not able to take care about the resource management then I better do not program. Vs. c#, where there are hidden traps ... I need to take care e.g. wether sometng is IDisponable or not and so on. Case by case, not straight forward. Only my 2 cents.
It does not solve my Problem, but it answers my question
-
Munchies_Matt wrote:
Of course they can be, they are the same language,
They are no longer the same language. Haven't been the same for quite some time. And actually the first ANSI C++ spec made at least one change that made it different from C.
jschell wrote:
made at least one change that made it different from C.
Which was?
#SupportHeForShe Government can give you nothing but what it takes from somebody else. A government big enough to give you everything you want is big enough to take everything you've got, including your freedom.-Ezra Taft Benson You must accept 1 of 2 basic premises: Either we are alone in the universe or we are not alone. Either way, the implications are staggering!-Wernher von Braun
-
Super Lloyd wrote:
For one thing it has little to do with "object lifecyle"
Perhaps we have a different definition of that term. I have seen problems in both Java and C# where the developers failed to properly account for how an object 'ended'. They had no problem with when the object 'started' but they never considered how to specifically control how it 'ended'. Matter of fact I suspect that I have seen code that might have that sort of problem just in the past week.
Super Lloyd wrote:
FYI, unless you do interop, which is less and less common those days, the only reason for leak in C# is if your referenced by static variable or (though is the same thing in disguise) static event. Or maybe too many live threads which don't die (and all the memory they collect), could typically come from wait handle hanging forever...
Probably been at least 15 years since I worked in a system where at least some threads did not die. Often quite a few of them. At least in Java the server frameworks build that in extensively. Most problems I have seen have been because the developers did not even seem to be aware that they were working in a threaded environment. Java and C# both have complex thread pool support and the concept has existed since before C# existed. Might have existed before Java but I wasn't introduced to it until after I first used Java. For context I started with Java 1.1.4 (well before 1.4.)
ThreadPools have been around since long before java's inventor was a gleam in his grandfather's eye.
#SupportHeForShe Government can give you nothing but what it takes from somebody else. A government big enough to give you everything you want is big enough to take everything you've got, including your freedom.-Ezra Taft Benson You must accept 1 of 2 basic premises: Either we are alone in the universe or we are not alone. Either way, the implications are staggering!-Wernher von Braun
-
CodeWraith wrote:
I have also seen an entire team of Java guys sitting around a server that needed a reset every day because their opus hogged memory.
I was told by a number of unix admins that when C/C++ servers were the norm that was the standard operating procedure for any large enterprise. It was something that was planned for when putting the process together for a new server application. And it wasn't just for memory. File descriptors and sockets also needed to be cleaned up.
That's SOP for any data center. Mostly because of the OS, not the applications.
#SupportHeForShe Government can give you nothing but what it takes from somebody else. A government big enough to give you everything you want is big enough to take everything you've got, including your freedom.-Ezra Taft Benson You must accept 1 of 2 basic premises: Either we are alone in the universe or we are not alone. Either way, the implications are staggering!-Wernher von Braun
-
Munchies_Matt wrote:
Of course they can be, they are the same language,
They are no longer the same language. Haven't been the same for quite some time. And actually the first ANSI C++ spec made at least one change that made it different from C.
In respect of the errors you can make not freeing memory they are, which was the part of the language we were discussing.
-
jschell wrote:
made at least one change that made it different from C.
Which was?
#SupportHeForShe Government can give you nothing but what it takes from somebody else. A government big enough to give you everything you want is big enough to take everything you've got, including your freedom.-Ezra Taft Benson You must accept 1 of 2 basic premises: Either we are alone in the universe or we are not alone. Either way, the implications are staggering!-Wernher von Braun
added plus plus to name to start with
-
:-D I should say that I had "early retired" back when auto_ptr was the new k3wl thing. I understand that after that, there have been some other features added that addressed issues that auto_ptr did not completely fix. I wonder if the state of the language is such that there are no more issues with memory leaks - which would indicate to me that no one need bother with calling delete anymore, basically making C++ like its managed cousin C#. Or am I missing something?
-
added plus plus to name to start with
ha ha not
#SupportHeForShe Government can give you nothing but what it takes from somebody else. A government big enough to give you everything you want is big enough to take everything you've got, including your freedom.-Ezra Taft Benson You must accept 1 of 2 basic premises: Either we are alone in the universe or we are not alone. Either way, the implications are staggering!-Wernher von Braun
-
jschell wrote:
made at least one change that made it different from C.
Which was?
#SupportHeForShe Government can give you nothing but what it takes from somebody else. A government big enough to give you everything you want is big enough to take everything you've got, including your freedom.-Ezra Taft Benson You must accept 1 of 2 basic premises: Either we are alone in the universe or we are not alone. Either way, the implications are staggering!-Wernher von Braun