Why do people use macros to declare namespaces in c++?

Compatibility with compilers that don’t support namespaces.

It's useful for compatibility, testing, etc. For example you may want to make use of inline namespaces but also support compilers that don't have that feature, yet, so you make it possible to disable it via the preprocessor.

LLVM itself also adds some other macros on top of the namespace declaration, making the macro shorter and avoiding repetition. See its definition in this file

Qt optionally uses a namespace to avoid linker errors in certain configurations. If Qt started today, they'd probably always use a namespace (and some inline namespaces on top) but it's a pretty old library with high regards for backwards compatibility, so they can't change now.

It's for ABI versioning. See the definitions of those macros.

If for some reason you need to use two versions of the same library in a single binary, you can change the namespace of one copy

BTW, don't look to standard library implementations for code that can be used in reasonable user programs. Remember that the standard library may make non-portable assumptions about the compiler it's used with, and is allowed to use otherwise-reserved identifiers (such as the macro name mentioned in this question).

Imagine you are an LLVM standard library developer, working with two standard C++ libraries. The std one, and an experimental experimentalstd one.

The LIBCPPBEGINNAMESPACESTD can be used to specify whether that part of the code is in the std namespace, or the experimentalstd namespace.

Gives you a chance to change the namespace in the remote possibility of a clash. I use BLAHBLAHNAMESPACEBEGIN and BLAHBLAHNAMESPACEEND, I like the readability.

One reason may be C compatibility or the programmer is coming from C background.