This may be an obsolete improvement when considered in light of #3145 and #3196, but I've put together modified versions of the make.bat.new_t and ~~~make.bat_t~~~ Jinja templates that enable handling of multiple make targets and (AFAIK) the full suite of sphinx-build command line options.

It runs through all of the arguments passed to make.bat and splits them into %TARGETS% and %OPTIONS%, then re-runs make.bat once for each target, passing in all of the %OPTIONS% that were provided. This makes it somewhat inefficient if, say, -E is passed, but I'm not sure how the behavior could be reasonably implemented otherwise. The script performs how I expect it to in a few minutes of poking at different valid and invalid invocation cases (Windows 7 64-bit, py3.5.1, all other packages those installed by default after a pip install -e . invocation in the repository root). The %OPTIONS% handling accounts (AFAICT properly) for the subset of options that take arguments.

Is this enhancement of interest to the project?

If so, I didn't see anywhere in the test suite where the commandline/make utilities were tested. If I missed something, please advise where to put any tests; or, if I should create a new suite for some tests, please advise of that.

Further: I'm not 100% sure about Vista / XP / 2k compatibility -- would I need to confirm any of those?

Finally: I've made these changes in a branch off of stable for now; please advise if I should rebase to master.

Thanks!