I think it would make sense to also change from --disable-zmq to --enable-zmq (to fit the current style --enable-wallet etc.)

I had the same feeling when I tested --enable-bench... But: --enable-tests is also enabled by default and the option is described as --enable... so I left it as it is now.

Looks like I misread what you wrote.

I personally prefer --disable-feature documented, when it is enabled by default. So I won't change anything here...

If it's enabled by default, the feature logically requires a --disable-foo. And the opposite, for anything disabled by default. (and as seen in #6733 and elsewhere, features should be built, even if runtime-disabled)

Right - the reasoning would be thus: you never have to pass --enable-zmq, because if the relevant library and version is installed, it picks it up automatically.
So documenting --disable-zmq makes sense.

BTW: if we want to be consistent, we should be consistent, looking at configure output, there are more options that start with uppercase, probably should adapt them as well:

  --with-boost-libdir=LIB_DIR
                          Force given directory for boost libraries. Note that
  --with-libressl         Build with system LibreSSL (default is no;
                          DANGEROUS; NOT SUPPORTED)

But re: libressl see also #6732

Boost one is an upstream issue in http://www.gnu.org/software/autoconf-archive/ax_boost_base.html - I do not want to divert.

Ad enable/disable rewrites:

  --enable-shared[=PKGS]  build shared libraries [default=yes]
  --enable-static[=PKGS]  build static libraries [default=yes]
  --enable-fast-install[=PKGS]
                          optimize for fast installation [default=yes]
  --enable-wallet         enable wallet (default is yes)
  --enable-tests          compile tests (default is yes)
  --enable-hardening      attempt to harden the resulting executables (default
                          is yes)
  --enable-ccache         use ccache for building (default is yes if ccache is
                          found)

This is a quick scan of the current configure here.

OK, fair enough, thanks for checking. Just want to avoid zillions of pulls that change the case of a single character, so if we could have done multiple at once it would be nice. But yes the boost one has to be changed upstream...

Nod - a pull request to change them all at once