No. Everything is a finite programmable automaton, the so-called "wired machine", and vulnerabilities in image parsers are also not uncommon.
I was talking purely about file format neutrality. The brain kept spinning for 5 mins on "wired machines" until I realized it was a typo. Yes I agree with you. But I still highly doubt if there will be ever one purely in the decoder part of the pipeline rather than processing/editing/post processing parts. The format parsing is already defined for decades right? Of course, some new crooked library will somehow manage to come up with a bug even in the decoder but the mature ones? The question is that the format of these 2 files: jpeg, png on disk is already simple, the parsing has been done by thousands of different projects with varying differences from copy paste to big or small changes and this has happened for decades. I am just purely talking about decoding, are such possible in good libraries if you follow the formal specification refined over the decades? It can be a chain of different parts that involve the decoding part but I still cannot see how good libraries have such bugs purely if let's hypothetically separate the decoder into an individual logic. But yes I agree that vulnerabilities in 'image processors', 'text editors' are not uncommon.
