Question

I am editing a program that uses a RandomAccessFile object, and I want to come up with my own RandomAccessFile class that uses a different source for the data other than a file object (it’s an Amazon webservices S3 object, but that’s irrelevant)

I want to basically make a class called RandomAccessS3 that has RandomAccessFile as its superclass, so I can simply say

RandomAccessFile raf = new RandomAccessS3();

and therefore keep the existing code the same. I will simply override every method in RandomAccessFile.

The problem is that in the subclass RandomAccessS3’s constructor I am forced to call RandomAccessFile’s constructor using super(file, mode) which takes as parameter a filename, and throws an error and dies if the file is invalid.

I can’t surround the super() call with a try/catch block because super is required to be the first line in the constructor. I could supply a dummy file, but I don’t want to force the user to do that. Is there any simple way around this?

Thanks!

Answer 1

There’s no simple way around this. The constructor is there to make sure that an instance is in a good/valid state, and part of the contract of RandomAccessFile is that an instance is linked to a valid file.

As a huge hack, I suppose that you could create a temp file, just to satisfy the RandomAccessFile constructor.

A better approach probably is to create a new interface and update your references from RandomAccessFile to the new interface. Then, implement the interface twice, once using RandomAccessFile and once with S3 calls.