7 ms·

>Tromp's entry is amazing. It really is, which leads me to wonder how difficult it would be, to deobfuscate this into code that can be understood by the averag

by wamatt 14y ago

>Tromp's entry is amazing. It really is, which leads me to wonder how difficult it would be, to deobfuscate this into code that can be understood by the average programmer? At least in terms of being able to say what it does, without being told. While most would probably regard security through obscurity as bad practice, perhaps there are practical uses in certain contexts.

gizmo686 14y ago
When you are talking about mid to high level security, you assume that your adversary is not an average programmer, but rather a train security analasyst. A major part of that field is looking finding vulnurabilities in machine code. Based on the statement "Obfuscation is due entirely to conciseness" in the author's hint, it sound like he did not use any techniques that would have a significant effect on the generated machine code.
astangl 14y ago
John asked me to review his entry prior to submission, and we had some ongoing discussion as he worked through his revisions. He initially pointed me to a "minimally obfuscated" version to analyze. I'd say it took me a good hour or so to feel like I had a pretty good understanding of its workings, and even then I still wasn't clear on some details of the I/O model and a couple other things. Most of the evolution from that point was finding ways to reduce the code size, then finally reformatting into a lambda. I credit John for not succumbing to gratuitous obfuscation, nor gratuitous use of macros to shave bytes. I suggested to him at one point to consider reducing the entire program to a single loop. He didn't think it would be shorter and pointed out that he had made a similar transformation in an earlier IOCCC entry of his, his infamous maze program, so I believed him and didn't go through the effort to verify for myself.

7 ms·

>Tromp's entry is amazing. It really is, which leads me to wonder how difficult it would be, to deobfuscate this into code that can be understood by the averag

by wamatt 14y ago

gizmo686 14y ago
When you are talking about mid to high level security, you assume that your adversary is not an average programmer, but rather a train security analasyst. A major part of that field is looking finding vulnurabilities in machine code. Based on the statement "Obfuscation is due entirely to conciseness" in the author's hint, it sound like he did not use any techniques that would have a significant effect on the generated machine code.
astangl 14y ago
John asked me to review his entry prior to submission, and we had some ongoing discussion as he worked through his revisions. He initially pointed me to a "minimally obfuscated" version to analyze. I'd say it took me a good hour or so to feel like I had a pretty good understanding of its workings, and even then I still wasn't clear on some details of the I/O model and a couple other things. Most of the evolution from that point was finding ways to reduce the code size, then finally reformatting into a lambda. I credit John for not succumbing to gratuitous obfuscation, nor gratuitous use of macros to shave bytes. I suggested to him at one point to consider reducing the entire program to a single loop. He didn't think it would be shorter and pointed out that he had made a similar transformation in an earlier IOCCC entry of his, his infamous maze program, so I believed him and didn't go through the effort to verify for myself.