Re:How C++ Resolves a Function Call
by WalterBright on Tuesday March 16, 2021 @04:17PM
(#26482253)
I implemented a printf format checker for D, which checks that the argument types are compatible with the format specifiers. (This is really a nice feature, I should have done it long ago.)
To be successful, it had to correctly deal with every nuance of the printf spec. cppreference got a couple of these incorrect wrt to modifiers. Not something a routine user would notice, but a pedantic one would. (Sorry, I don't remember exactly what the mistakes were.)
I still use cppreference because it is so dang convenient. But I don't rely on it when perfection is required.
It's really too bad that copyright forces everyone writing a manual to rewrite & rephrase things. I wish the C++ Standard was online.
One online reference I dearly like is https://www.felixcloutier.com/x86/index.html which has saved me so much time. But it was created by scanning the actual reference manuals, so barring scanning errors, it is exactly correct. I gave up using reformulations of the CPU instruction set long ago, they had too many mistakes.
Re:How C++ Resolves a Function Call
by jeffbee on Tuesday March 16, 2021 @04:55PM
(#26482848)
You might be interested in the EXEgesis project, which is also a machine interpretation of x86 documentation, but which includes several patches for unfixed errata therein.
To be successful, it had to correctly deal with every nuance of the printf spec. cppreference got a couple of these incorrect wrt to modifiers. Not something a routine user would notice, but a pedantic one would. (Sorry, I don't remember exactly what the mistakes were.)
I still use cppreference because it is so dang convenient. But I don't rely on it when perfection is required.
It's really too bad that copyright forces everyone writing a manual to rewrite & rephrase things. I wish the C++ Standard was online.
One online reference I dearly like is https://www.felixcloutier.com/x86/index.html which has saved me so much time. But it was created by scanning the actual reference manuals, so barring scanning errors, it is exactly correct. I gave up using reformulations of the CPU instruction set long ago, they had too many mistakes.