How to understand the reliability provided by AppDomain Isolation?

Question

4.00/5 (3 votes)

See more:

I got confused about the description of AppDomain in MSDN.[^]

It's said:
Faults in one application cannot affect other applications. Because type-safe code cannot cause memory faults, using application domains ensures that code running in one domain cannot affect other applications in the process.

But any unhandled exception in any thread, no matter which AppDomain the code runs in, will cause the entire process crash. So how to understand the reliability provided by AppDomain?

Posted 3-Aug-12 9:42am

xTerryx

Add a Solution

Comments

[no name] 3-Aug-12 15:45pm

The answer to your question is right there in your statement, "Faults in one application cannot affect other applications"

xTerryx 3-Aug-12 16:06pm

but if there is any unhandled exception, no matter which AppDomain the exception is thrown, will cause the entire process crash. So does it have conflicts with "Faults in one application cannot affect other applications"?

[no name] 3-Aug-12 16:10pm

What does that have to do with anything? So your app crashes. It's not going to affect other applications.

xTerryx 3-Aug-12 16:17pm

I mean, there are multiple AppDomains, a unhandled excepion happens in one AppDomain, other AppDomains will also crash. so it's not reliable, righ?

[no name] 3-Aug-12 16:18pm

No. Where do you get that? Thread != App domain

xTerryx 3-Aug-12 16:24pm

yes, thread != app domain. So "Faults in one application" does not mean "eception in app domain"? and what does it exactly mean?

[no name] 3-Aug-12 16:28pm

It's really simple. An exception in your application is not going to affect another application that is running in its own app domain. So it means exactly what you quoted.

xTerryx 3-Aug-12 16:32pm

so another application is another process, right? what has to do with app domain? processes are already isolated. one process crashes, another is still alive.

[no name] 3-Aug-12 16:35pm

That's what an app domain is, "an isolated environment where applications execute"

Sergey Alexandrovich Kryukov 3-Aug-12 16:52pm

OK, you can see my answer I've finally written, and perhaps stop this discussion.
Looking at OP's arguments (and your arguments are quite valid), I don't think further discussion is useful enough... :-)
--SA

Sergey Alexandrovich Kryukov 3-Aug-12 16:42pm

What is "crashes"? One of your problems is using this notion.
--SA

Sergey Alexandrovich Kryukov 3-Aug-12 16:41pm

You are using communication disruption as the argument. Do you want to trick the opponent, or yourself? It is unrelated to your own topic.
First of all, there is no such thing as crash. An exception in one AppDomain 1) can propagate to another one, on the same thread, 2) this is not unreliable, as always with exceptions.

Exceptions are related to stack, and AppDomains can share the same thread, and, hence, the stack.

--SA

ZurdoDev 3-Aug-12 16:49pm

When the documentation says that the process will crash it is referring to the app domain. Each app domain creates a new process. I think that is where the confusion is.

Sergey Alexandrovich Kryukov 3-Aug-12 16:53pm

Sorry, this is not true. An application domain does not create a process. Several application domains lives in one process (and there are many cases where it's very important to use them, such as reloadable plug-ins); please see my answer.
--SA

ZurdoDev 3-Aug-12 16:56pm

I was thinking of App Domains in IIS.

Sergey Alexandrovich Kryukov 3-Aug-12 17:02pm

I don't see anything about IIS... all issues look like the issues of .NET AppDomain.
--SA

ZurdoDev 3-Aug-12 17:04pm

IIS also has something called App Domains.

Sergey Alexandrovich Kryukov 3-Aug-12 16:38pm

What an empty-minded question! If you don't understand some more basic things, this is quite hard to explain. Well, imagine that some local variables in method would be visible and accessible from other methods. Could it lead to some disaster in the development? Same thing is process isolation and application domain isolation. The memory spaces of different application domains are totally unrelated. Cannot see a value of it..? Then keep practicing...
--SA

xTerryx 3-Aug-12 20:56pm

That's isolation of memory. One AppDomain cannot access the memory of another directly and they have different assemblies loaded, in another word, code in different AppDomain are isolated. But what I really want to know is what does the statement "Faults in one application cannot affect other applications" mean? "Faults" -> "Exception"? and "application" -> "AppDomain"?

1 solution

Add a Solution

Add your solution here

Treat my content as plain text, not as HTML

Preview 0

…

Existing Members

Sign in to your account

...or Join us

Download, Vote, Comment, Publish.

Your Email
Password
Forgot your password?

Your Email
This email is in use. Do you need your password?
Optional Password

I have read and agree to the Terms of Service and Privacy Policy
Please subscribe me to the CodeProject newsletters

When answering a question please:

Read the question carefully.
Understand that English isn't everyone's first language so be lenient of bad spelling and grammar.
If a question is poorly phrased then either ask for clarification, ignore it, or edit the question and fix the problem. Insults are not welcome.
Don't tell someone to read the manual. Chances are they have and don't get it. Provide an answer or move on to the next question.

Let's work to help developers, not make them feel stupid.

This content, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)

Sergey Alexandrovich Kryukov · Accepted Answer · 2012-08-03T10:50:00

You main problem is that you are using inadequate notion of "crash".

In essence, having unhandled exception in an application is strictly equivalent to exit from the entry point method, if you think about it. You have some exception propagation along the stack of some thread. When the instruction flow jumps to the top of the stack with exception mechanism, and top-level try-catch block was not used, the application has nothing to do but terminate the thread. If it was the only thread in the process, the process terminates. But even termination of the process does not always happen, because it can be some other thread, not the one of the entry point.

Now, Application Domain does not change this picture much. Actually, Application Domains can share threads and hence, the stacks of them. So, you can make a call in one Application Domain under try-catch, have an exception thrown in another Application Domain, not handled (which is actually good). In this case, the exception will propagate back to the calling Application Domain. And, optionally, caught there.

This does not reduce reliability even a bit, but, if exception is properly handled on top of stack of each thread (no matter which Application Domain), can greatly improve it.

—SA