Consider making a new release

[ryandesign]

umem 1.0.1 does not build on Apple Silicon systems.

The config.guess and config.sub included in the tarball are too old. The error is:

checking build system type... configure: error: /bin/sh ./config.sub -apple-darwin20.5.0 failed

This is solved by getting new config.guess and config.sub files from automake 1.16.3 or later.

After fixing that, the next error is:

In file included from init_lib.c:37:
In file included from ./umem_base.h:32:
In file included from ./umem_impl.h:49:
./sol_compat.h:135:12: error: implicit declaration of function 'ec_atomic_cas' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
  } while (ec_atomic_cas(mem, last+1, last) != last);
           ^
./sol_compat.h:135:12: note: did you mean 'ec_atomic_inc'?
./sol_compat.h:130:22: note: 'ec_atomic_inc' declared here
static INLINE uint_t ec_atomic_inc(uint_t *mem)
                     ^
1 error generated.

This was solved by cbaa083.

Might you consider making a new release to address these and any other issues that have been fixed since 1.0.1 was released in 2009?

[wez]

The people that were working on this way back in ~2010 aren't working on this any more--it was more of a Message Systems project than an OmniTI project, and both of those companies have since been acquired and pivoted somewhat. I wouldn't expect to see anything else change in this repo.

If you're looking for a spiritual successor, I'd suggest looking at: https://github.com/backtrace-labs/slitter
(with more background here: https://engineering.backtrace.io/2021-08-04-slitter-a-slab-allocator-that-trusts-but-verifies/)
which is a modern alternative implemented by the fine folks at Backtrace.io.