As I said, you can also win the original challenge with this method. The original challenge is not fully specified. In particular it is not specified how the decompressor is to be encoded. There are encodings for the decompressor for which the strategy that I described does work, and encodings for which it does not work.

The interesting attribute of this method is that it does genuinely compress the file: it doesn't store data in hidden attributes like the method of the OP (which stores information in the file lengths).