Nguyễn Thái Ngọc Duy <pclouds@xxxxxxxxx> writes: > Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@xxxxxxxxx> > --- > apply.c | 34 +++++++++++++++++++--------------- > 1 file changed, 19 insertions(+), 15 deletions(-) Until all the codepaths that reach these callsites to index_* functions from cmd_main() are converted to pass "struct index_state *" throughout, a step like this patch does fundamentally changes nothing. The only two issues worth addressing in this area still remain. All the lines we see here still depend on the existence of "the_index" instance, and they can only operate on that single "the_index" instance and nothing else. It is true that the dependency has been made more explicit, but it already is explicit (to see them you just say "no-the-index-macros" and see what fails to compile). I see others are enthused with this series, but seeing the changes like these I am not all that impressed. Would it become hard to review if we combine this step *and* the next logical step (i.e. pass the "struct index_state" throughout the callflow) into a single patch? The functions in apply.c are fairly well isolated and there won't be all that heavy interaction with the outside world if we convert this file (and it alone) without touchning the other files. If a division in that direction is possible, it may make a better orgainzation (i.e. instead of doing whole-tree superficial conversion that needs to be fixed up again later, do a deep full conversion on selected files before going on to next set of files).
