These two commits make fetch quite a bit faster when dealing with
large numbers of refs. Unfortunately I've lost the exact figures, but
my silly test respository (50000 tags, 1 branch) went from something
like 30s for a "fetch --tags" that did nothing, to about 5.
Julian Phillips (2):
remote: Make ref_remove_duplicates faster for large numbers of refs
fetch: Speed up fetch of large numbers of refs
builtin-fetch.c | 17 ++++++++++++++---
remote.c | 39 +++++++++++++++++++++------------------
2 files changed, 35 insertions(+), 21 deletions(-)