"That makes everything slow, inefficient, and widely dangerous."
There nothing faster and more efficient than building C programs. I also not sure what is dangerous in having libraries. C++ is quite different though.
back
2 comments
Of course there is. Raw machine code is the gold standard, and everything else is an attempt to achieve _something_ at the cost of performance, C included, and that's even when considering whole-program optimization and ignoring the overhead introduced by libraries. Other languages with better semantics frequently outperform C (slightly) because the compiler is able to assume more things about the data and instructions being manipulated, generating tighter optimizations.
I was talking about building code not run-time. But regarding run-time, no other language does outperform C in practice, although your argument about "better semantics" has some grain of truth in it, it does not apply to any existing language I know of - at least not to Rust which is in practice for the most part still slower than C.
ODR violations are very easy to trigger unless you build the whole thing from source, and are ill-formed, no diagnostic required (worse than UB).
Neither "ODR violations" nor IFNDR exist in C. Incompatibility across translation units can cause undefined behavior in C, but this can easily be avoided.
C simply has less wording for it because less work has been put into it.
The same problems exist.
The ODR problem is much more benign in C. Undefined behavior at translation time (~ IFNDR) still exists in C but for C2y we have removed most of it already.
You can't fundamentally solve the issue of what happens if you call a function in another TU that takes a T but the caller and the callee have a different definition of T.
Whether you call that IFNDR or UB doesn't make much of a difference.
C++ mitigates that issue with its mangling (which checks the type name is the same), Rust goes the extra mile and puts a hash of the whole definition of the arguments in the symbol name.
C has the most unsafe solution (no mitigation at all).
C++ ODR requires different definition to consist of the same tokens, and if not the the program is IFNDR. Name mangling catches some stuff, but this becomes less relevant today with more things being generated via templates from headers.
In C, it is UB when the types are not compatible, which is more robust. In practice it also easy to avoid with the same solution as in C++, i.e. there is a single header which declares the object. But even if not, tooling can check consistency across TU it is just not required by the ISO standard (which Rust does not have, so the comparison makes no sense). In practice, with GCC a LTO build detects inconsistencies.
Different parts of the build seeing inconsistent definitions of the same name is a clear consequence of building things piecemeal rather than as a single project -- which is precisely the problem I described higher up in this thread.
Things being built piecemeal also likely won't be using LTO (even if fat LTO allows this, no static library packages in a distro are built with it).