Junio C Hamano [off-list ref] writes:

Karthik Nayak [off-list ref] writes:

quoted

This explains for why 'broken' must use a subprocess, but there is
nothing stopping 'dirty' from also using a subprocess, right? It
currently uses an in-process index refresh but it _could_ be a
subprocess too.

Correct, except that it does not make sense to do any and all things
that you _could_ do.  So...

Well, In this context, I think there is some merit though. There are two
blocks of code `--broken` and `--dirty` one after the other which both
need to refresh the index. With this patch, 'broken' will use a child
process to do so while 'dirty' will use `refresh_index(...)`. To someone
reading the code it would seem a bit confusing. I agree there is no
merit in using a child process in 'dirty' by itself. But I also think we
should leave a comment there for readers to understand the distinction.