Re: When to explicitly check permission
From: Nicole Calinoiu (calinoiu)
Date: 10/29/04
- Next message: Nicole Calinoiu: "Re: Check for permission does not work"
- Previous message: David Seimour: "Using COM objets from User Control embedded in Internet Explorer"
- In reply to: Andreas Håkansson: "When to explicitly check permission"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: Fri, 29 Oct 2004 12:13:31 -0400
"Andreas Håkansson" <andreas.hakansson@nospam.com> wrote in message
news:e0H1hJcvEHA.1988@TK2MSFTNGP12.phx.gbl...
> When should you explicitly check for correct permission? Perhaps this
> sounds
> strang, I know, but stick with me for a moment. For example the
> GetCurrentDirectory
> method on the Directory class throws an UnauthorizedAccessException if the
> caller doesn't have access to the I/O operation, so there wouldn't be any
> point
> in doing a Demand call on a FileIOPermission object and possibly trap a
> SecurityException instead would there?
The UnauthorizedAccessException is thrown when the operating system denies
access, not when .NET denies access due to CAS permissions. However, the
Directory.GetCurrentDirectory method will implement a demand for
FileIOPermission.PathDiscovery for the target directory (assuming the
Windows API call doesn't fail first), so there's probably not much point in
implementing your own demand.
> The Exists method on the File class performs a call to the Demand method
> on a
> FileIOPermission object interally, but never rethrows the
> SecurityException (among
> others) which can occure internally in the method, but instead returns
> False - so why
> should I do an explicit check my self before I call the method? Or
> shouldn't I ? =)
If the behaviour as implemented is OK with you, there's no point in
performing a preliminary demand for the same permission.
> The Framework is litered with scenarios like these. A few places I have
> used it is with
> the SQLClientPermission and SecurityPermission (to be sure I could call
> into unmanged
> code - COM Interop scenario), but I would like some more generall
> guidelines for this
> kind of stuff to improve the overall quality of my code.
There's really only one scenario in which you might typically want to
implement a demand of your own for a permission that's already being
demanded by a method that you're calling. This is when you're going to loop
through many calls of the target method, and you want to avoid the full
stack walk in each iteration of the loop. You could instead use a sequence
like the following: demand -> assert -> loop -> revert the assertion. Of
course, the usual assertion caveats apply.
>
> Cheers! =)
>
> //Andreas
>
>
>
- Next message: Nicole Calinoiu: "Re: Check for permission does not work"
- Previous message: David Seimour: "Using COM objets from User Control embedded in Internet Explorer"
- In reply to: Andreas Håkansson: "When to explicitly check permission"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|
