Comments on: Compiler writing: The career path to World domination http://shape-of-code.coding-guidelines.com/2009/11/07/compiler-writing-the-career-path-to-world-domination/ Sun, 12 Feb 2012 22:01:48 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: yugr http://shape-of-code.coding-guidelines.com/2009/11/07/compiler-writing-the-career-path-to-world-domination/comment-page-1/#comment-894 yugr Fri, 25 Dec 2009 21:38:03 +0000 http://shape-of-code.coding-guidelines.com/?p=125#comment-894 Well, CIA or NSA (or their equivalents in other countries) may invest a few man-years to build a very basic compiler just for a purpose of gcc validation. Well, CIA or NSA (or their equivalents in other countries) may invest a few man-years to build a very basic compiler just for a purpose of gcc validation.

]]> By: Derek-Jones http://shape-of-code.coding-guidelines.com/2009/11/07/compiler-writing-the-career-path-to-world-domination/comment-page-1/#comment-744 Derek-Jones Wed, 18 Nov 2009 02:23:17 +0000 http://shape-of-code.coding-guidelines.com/?p=125#comment-744 David Wheeler's thesis blog post is linked to through "a recent PhD thesis". I was a bit disappointed that this thesis wasted so much space on mathematical proofs of little practical relevance and less space on realistic verification issues. I don't think it is realistic to assume the existence of a trusted compiler. We should assume that all compilers contain backdoors. Analysing in more detail how actual implementations might function would provide data that could be used to estimate the probability that two independently developed compilers would use the same backdoor techniques and thus make it lot less likely that one could be used to verify the other (making the worse case scenario that all compilers contain backdoors). It could also provide useful insight into how source code evolution impacts backdoor propagation success rates. David Wheeler’s thesis blog post is linked to through “a recent PhD thesis”. I was a bit disappointed that this thesis wasted so much space on mathematical proofs of little practical relevance and less space on realistic verification issues.

I don’t think it is realistic to assume the existence of a trusted compiler. We should assume that all compilers contain backdoors.

Analysing in more detail how actual implementations might function would provide data that could be used to estimate the probability that two independently developed compilers would use the same backdoor techniques and thus make it lot less likely that one could be used to verify the other (making the worse case scenario that all compilers contain backdoors). It could also provide useful insight into how source code evolution impacts backdoor propagation success rates.

]]> By: Tony Finch http://shape-of-code.coding-guidelines.com/2009/11/07/compiler-writing-the-career-path-to-world-domination/comment-page-1/#comment-742 Tony Finch Tue, 17 Nov 2009 21:27:17 +0000 http://shape-of-code.coding-guidelines.com/?p=125#comment-742 This thesis is relevant: Fully countering \Trusting Trust\ by diverse double-compiling. http://www.dwheeler.com/blog/2009/11/02/#trusting-trust-dissertation This thesis is relevant: Fully countering \Trusting Trust\ by diverse double-compiling.

http://www.dwheeler.com/blog/2009/11/02/#trusting-trust-dissertation

]]>