"John Cai via GitGitGadget" [off-list ref] writes:

This commits adds a flush function pointer as a new member of
reftable_writer because we are not sure that the first argument to the
*write function pointer always contains a file descriptor. In the case of
strbuf_add_void, the first argument is a buffer. This way, we can pass
in a corresponding flush function that knows how to flush depending on
which writer is being used.

A comment and a half.

 * Can't the new "how to flush" go to the write-option structure?
   If you represent "no flush" as a NULL pointer in the flush member,
   most of the changes to the _test files can go, no?

 * For a function

	int func(int ac, char **av);

   a literal pointer to it can legally be written as either

	int (*funcp)(int, char **) = &func;
	int (*funcp)(int, char **) = func;

   but it is my understanding that this codebase prefers the latter,
   a tradition which goes back to 2005 when Linus was still writing
   a lot of code, i.e. the identifier that is the name of the
   function, without & in front.