As I’ve said below, if you have a large, old code base which has to build for a lot of platforms, migrating to a new compiler is a serious undertaking.
Moving a million lines of source code from one compiler to another is non-trivial.
At work, we compile with very conservative settings (-Wall -Werror, among others), and we still have trouble moving from one version of GCC or GNU ld to another, let alone move to another toolchain altogether.
LLVM also has less battle hardened backends for embedded architectures (MIPS, PowerPC), and lacks the level of commercial support MontaVista and others provide for GCC (and packages built with GCC…).
Migrating to clang++/LLVM would require a company to solve and maintain workarounds for certain compiler issues themselves, which is a significant investment.
This is really good news for someone who programs in C++ every day :) Would love to move away from GCC.
Is there something big stopping you from moving away from gcc?
I’m looking forward to checking out the
-Rflag, myself.As I’ve said below, if you have a large, old code base which has to build for a lot of platforms, migrating to a new compiler is a serious undertaking.
Yep … been there. That’s why I was curious what the issue was. :)
[Comment removed by author]
Moving a million lines of source code from one compiler to another is non-trivial.
At work, we compile with very conservative settings (-Wall -Werror, among others), and we still have trouble moving from one version of GCC or GNU ld to another, let alone move to another toolchain altogether.
LLVM also has less battle hardened backends for embedded architectures (MIPS, PowerPC), and lacks the level of commercial support MontaVista and others provide for GCC (and packages built with GCC…).
Migrating to clang++/LLVM would require a company to solve and maintain workarounds for certain compiler issues themselves, which is a significant investment.