|
[background info]
For a project I'm working on I'm looking at using run-time compilation of regular cs files into an appdomain separate from the running assembly (to disallow filesystem and network access) and reference the objects the files define. This is becaues there will be alot of additions to the group of objects those files represent, while the core of the program should remain the same. I'm using the Microsoft.CSharp.CSharpCodeProvider object to do the compilation.
[/background info]
Now, I would like to expose the running assembly's public classes and interfaces so that the run-time compiled code may reference and make use of them. First and foremost it's for exposing an interface that every object compiled this way must implement. I am not completely sure how to accomplish this, even after trawling through intellisense and msdn.
Anybody got any clues for me?If I am unclear anywhere or additional info is needed, just ask!
|
|
|
|
|
Create an interface for the public properties of your running assembly, and an object implementing that interface and filled properly. Let's call it the IHost interface and the Host class.
Now you compile your dynamic code, and instantiate its objects. Those objects should implement another interface with a SetHost(IHost myHost) method. Call it with the Host object as a parameter, and now your dynamic objects can access all the properties required.
|
|
|
|
|
Your best bet would be to compile your model layer into a separate dll which can be referenced by both your main app as well as new assembly. You can reference the dll using the CompilerParameters class. I looked into this a while ago, trying to create a C# scripting engine for our app, but in the end I went with IronPython as the examples and documentation were far more widespread. The main problem I came up against was referencing Instances of objects that were created in the main AppDomain. I gave up, and I don't do that very often. Good luck, and if you get it working I would really like to see an article on how you did it. I still think a C# scripting engine would be better than Python, because we all know c# here.
|
|
|
|
|
I'm leaning towards this, and I've been pointed in a similar direction elsewhere as well. A test project let me have a simple version up and running fairly quickly,though the big stumbling block came when trying to restrict permissions for the compiled assembly. I can run the code in a restrictive, separate appdomain, but I can't do it without giving Unrestricted FileIOPermission to that domain, which kinda defeats the point I was going for. It almost seems like it needs Unrestricted FileIO in order to load the exposed assembly?
|
|
|
|
|
I also managed to get test projects running. These were really simple projects that weren't much good for real life. I did manage to get away from using System.IO by compiling the code on-the-fly using the CompileAssemblyFromSource method rather than saving to a file and compiling the source file, but I wasn't really looking at restricting access to anything. Basically I was looking at a Database management app that could handle all three database types we use(MySql, SQLServer and Access(  )), and all the users of such an internal app would have been developers.
|
|
|
|
|
I have developed something similar to this. In my case, that application did not compile cs files at runtime, but I think the idea might be the same becouse it loaded external libraries at runtime, and could use objects defined into it. Here is, basically, what I did:
1. One library where the required interfaces where defined (just the interfaces).
2. Several libraries which referenced the "interface library" and implemented them as needed to achieve their own features.
3. One main application which referenced the "interface library" as well, and loaded the libraries described in point 2 at runtime (whenever it was required), and used objects of them through the intefaces.
This is the way that my main application could be extended with modules (or plugins).
Now, for the security issues you describe I think you might use Caspol[^] to decide the permissions you give to those libraries, but I am not sure if it would work... I have not made any test about this.
|
|
|
|
|
Whoop! Got to work, at least on a medium scale! I think part of what I was doing wrong in my last post was I accidentally set up TWO levels of restrictive appdomains, i.e. the main app set up a restrictive appdomain, then run the sandboxing code in that appdomain, which IN TURN created a restrictive appdomain and tried to load my intended sandboxed assembly into it (but failed because of the first level appdomain being restrictive)
So, to summarize:
A)Create a secondary Class Library project in the solution, let's call it "Exposed", which contains everything that the cs files will have access to.
B)Gather all the cs files I want to compile. The cs files reference "Exposed" in their code.
C)Use a CSharpCodeProvider to CompileAssemblyFromFile() those files into a class library dll-file saved in the appdir.Provide "Exposed.dll" as a reference. Provide compiler options "/target:library /out:*dll-name*.dll".
---The below steps are performed by a slightly customized Sandboxer object detailed here MSDN article[^]---
D)Create the appdomain, taking care to supply a proper PermissionSet (Containing only Execution permission in this case) as well as the StrongName of the current assembly to add it as a FullTrustAssembly. Also set the ApplicationBase (in the AppDomainSetup object) to something other than your main app base directory, for security reasons.
E)Use System.Activator to load a new instance of the Sandboxer class into the restrictive appdomain.
F)The new instance in the restrictive appdomain now invokes the wanted method in the compiled library, taking care to catch SecurityExceptions.
Of course, this can be improved. Currently, the restrictive appdomain is created EVERY TIME the compiled code is called, for instance.
modified on Tuesday, March 29, 2011 7:15 AM
|
|
|
|
|
Good that you got it working. Just a reminder that anything that's created in one app domain and referenced in another must be marshalable (simple types or inheriting from MarshalByRefObject), and you might have to create proxy classes for some things which inherit from something else (I did when I wrote the plugin system for my game client).
|
|
|
|
|
Hi guys ,
I have problem with sending message to serial port
I have arraylist called MessageTimeAssociation and in it's first index of every message I have the time which determine when I will send Message into serial Port.
ex. if the first index value =5 I should send message after 5 seconds
so I am variable xCounter and I increment it with timer but My problem if time of message not equal to the xCounter the condition not success so the message skipped.
So I want this condition wait execution while Xcounter match the value of time.
How i can do that ?
ArrayList MessageTimeAssociation = new ArrayList();
int xCounter = 0;
private void timer1_Tick(object sender, EventArgs e)
{
foreach (string[] msg in MessageTimeAssociation)
{
if ((msg[1] == (setListBoxIndex.ToString()) &$ (int.Parse(msg[0])) == xCounter))
{
ConnectToSerialPort();
int q = 0;
numberOfMessagesInSection = msg.Length;
for (q = 2; q < msg.Length; q++)
{
byte[] myTempRcv ={ byte.Parse(msg[q]) };
serialPort.Write(myTempRcv, 0, 1);
}
}
}
xCounter++; ;
}
|
|
|
|
|
Hi,
that is pretty confusing. Here are some ideas:
1.
please show your code inside PRE tags, not CODE tags. That will preserve formatting and indentation, making things much more readable.
2.
also show the declarations of the variables you are using.
3.
it looks as if each item in the array list is a string array, with some specific values at specific locations. That is not so good. You should define your own type (class or struct) and store instances thereof inside your collection; the collection could better be a generic list, ArrayList is archaic and has no advantage over a generic list.
4.
I am surprised you seem to have a "ConnectToSerialPort" inside a loop. I doubt that is correct.
5.
This is what I would do:
- keep the messages to be sent (each as an instance of some Message class) inside a generic list, which is ordered by increasing due datetime.
- have a timer ticking at an appropriate interval (e.g. 1 second), then checking the current time with the time of the first entry in the list; if current time greater equal due time, remove message from list and send it.
This means the tick method would never wait, and that is how it should be: event handlers should never wait, since they would immobilize the GUI while they wait.
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.
|
|
|
|
|
|
OK, you have taken care of my earlier #1 and #2. That still leaves #3, #4, and #5.
I also noticed something else, so:
6.
for (q = 2; q < msg.Length; q++) {
byte[] myTempRcv ={ byte.Parse(msg[q]) };
serialPort.Write(myTempRcv, 0, 1);
}
does not look good. Why would you send one byte at a time?
here is something much more efficient:
int length=msg.Length;
byte[] myTempRcv =new byte[length];
for (q = 2; q < length; q++) {
byte[q]=byte.Parse(msg[q]);
}
serialPort.Write(myTempRcv, 2, length-2);
BTW: I am not sure the indexes are all correct, [ADDED] in fact I'm not sure your original code is correct as I don't know what the intention is, what the original content of msg is, and what exactly should be written to the serial port. [/ADDED] If you need more help about this, please provide an example with actual numbers (in decimal, or in hexadecimal, whatever suits you).
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code. modified on Sunday, March 27, 2011 7:29 PM
|
|
|
|
|
Hi to codeproject world!
i have a project with sql server 2008 which has a server and 5 client.
[i mean that the project i have written is running on 6 pc's(one server and 5 client)]
now my problem is when the client1 is editting a record ( for exm:- Id:12 , from table1 )
other 4 client and sever should not have permission to edit that record.
hint :
i send Id to sql and get record and then i start to edit. at that time i am not connected to sql..
|
|
|
|
|
Flags.
Set a flag on each record that indicates whether it is being edited or not.
If the flag is 0 then editing allowed, if 1 then not.
Once the edited record is completed, reset the flag to 0.
Simples.
------------------------------------
I will never again mention that I was the poster of the One Millionth Lounge Post, nor that it was complete drivel. Dalek Dave
CCC Link[ ^]
Trolls[ ^]
|
|
|
|
|
THE Problem is that if client1 is editting the record and before ending the system of client1 restarts . So how can other edit that record.(in this solution this record cant be edit for ever....)
|
|
|
|
|
Thats an armageddon scenario.
Why would the system restart?
If it restarts without ending the edit, then the edit would be lost.
You could put a time limit on open edited items, if not processed or closed within, say, 10 minutes, it closes automatically and resets the flag to 0.
------------------------------------
I will never again mention that I was the poster of the One Millionth Lounge Post, nor that it was complete drivel. Dalek Dave
CCC Link[ ^]
Trolls[ ^]
|
|
|
|
|
Yeah, and the flag would never get reset. What he wants just isn't really feasible to do and still have fault tolerance. One possible cause of a "failure" would be if Windows shutdown while the record was 'checked out' for edit.
|
|
|
|
|
Hi everyone,
I'm also facing same kind of problem.
Any help would be appreciate.
|
|
|
|
|
Then read the entire thread and figure out that what you want isn't really possible. You simply have to do research on "database concurrency" and pick the best option that fits your requirements.
|
|
|
|
|
Optimism is a great virtue, but not for a software developer...
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.
|
|
|
|
|
I would suggest reposting this question in the Database forum; this has very little to do with C#. AFAIK you can lock database records to do what you want; but I don't know the technical details, I expect someone in the Database forum knows the answer.
I must get a clever new signature for 2011.
|
|
|
|
|
Good advice.
------------------------------------
I will never again mention that I was the poster of the One Millionth Lounge Post, nor that it was complete drivel. Dalek Dave
CCC Link[ ^]
Trolls[ ^]
|
|
|
|
|
As Richard suggested, this question suits better in the Database forum.
You can make use of with(rowlock) in your update statement. This will help you avoiding multiple edits on the same row. You can also make use of rowversion to do this. I don't know much about rowversion way but someone in database forum should be able help with this.
|
|
|
|
|
and once you're in the Database forum, have a look at this thread[^]; I think it relates to what you want.
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.
|
|
|
|
|
Hi,
look at my Code. In my code DB_Command.Transaction = DB_Connection.BeginTransaction(System.Data.IsolationLevel.Serializable); defines transaction level. It locks(for update/insert) the database until the command is committed. Please never use flags.
public void Execute(string querystring)
{
DB_Command = new OleDbCommand(querystring, DB_Connection);
DB_Command.Transaction = DB_Connection.BeginTransaction(System.Data.IsolationLevel.Serializable);
DB_Command.CommandType = System.Data.CommandType.Text;
int afr=DB_Command.ExecuteNonQuery();
DB_Command.Transaction.Commit();
}
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
