Question

I’ve been writing C code for quite some time but am very new to Objective-C and OO program design. I am having trouble wrapping my head around how to design my program and it’s classes — my standby procedural programming paradigm of passing an opaque pointer to everything just won’t die a quiet death. I am writing a simple program in Xcode4 in Objective-C and I have some questions on design and convention. I’ve designed two classes “PacketCapture” and “IPGeo”. The first is a light wrapper around libpcap with some added parsing logic and the second is a light wrapper around MaxMind’s GeoIP C API. In my main module, I instantiate the PacketCapture class and pass control to the packet capture loop:

int main(int argc, const char *argv[])
{
    @autoreleasepool
    {
        PacketCapture *_PacketCapture = [[PacketCapture alloc] init:"en1" withFilter:""];
        if (_PacketCapture == nil)
        {
            NSLog(@"Can't instantiate PacketCapture class");
            return (-1);
        }
        IPGeo *_IPGeo = [[IPGeo alloc] init:"/usr/local/share/GeoIP/GeoIPCity.dat"];
        if (_IPGeo == nil)
        {
            NSLog(@"Can't instantiate IPGeo class");
            return (-1);
        }

        NSLog(@"entering packet loop...");
        [_PacketCapture packetCapture];
    }
    return (0);
}

Inside of the packetCapture method a parsing function is called that chops up the packet and displays it. I need to use the GeoIP logic down inside the parsing code to lookup the IP addresses against the geo database. Therein lies my problem and general confusion.

My questions:

• Where should I instantiate the GeoIP class?

• Is having an infinite loop inside an object in a class bad OO design? Feels like it could be. Perhaps if I restructure this problem will be easily solved?

Thanks for any help.

Answer 1

Where should I instantiate the GeoIP class?

If PacketCapture requires an IPGeo object to function, I would recommend instantiating it within PacketCapture‘s -init method. If you need to configure the IPGeo object (by passing a different path when initialized, for example) before using it in PacketCapture, and the path to use is determined in a way PacketCapture shouldn’t be aware of, have the class responsible for managing the two (or, in your case, the main loop of the program) create an instance and pass it as an argument to a custom initializer method on the PacketCapture class.

A bit of nitpicking–your initializer for IPGeo doesn’t follow Objective-C conventions for expressive method names. -initWithPathToDatabase: would look a lot better in my opinion.

In a similar vein, you can add a -initWithIPGeo: method to your PacketCapture class.

Is having an infinite loop inside an object in a class bad OO design?

If this is an iOS or Cocoa application, then definitely yes. If it’s a command line tool, then maybe. Furthermore, if that infinite loop is being run on the main thread, it will block all UI elements, which will make your program look unresponsive.

A better alternative would be to use the protocol/delegate pattern, a staple of Objective-C programs. Create a protocol with something like the following method:

- (void)packetCapture:(PacketCapture *)packetCapture didReceiveDataFromIPAddress:(NSString *)ipAddress;

Then create a delegate object that implements this protocol and performs an action on receiving the packets. If you’re making a command line tool, you can instantiate that delegate and wait to receive messages.