From: Junio C Hamano Date: Thu, 11 Jan 2007 00:34:11 GMT Subject: Re: [PATCH] Speedup recursive by flushing index only once for all entries Message-ID: <7v4pqy8kqk.fsf@assigned-by-dhcp.cox.net> In-Reply-To: <81b0412b0701101507n764aed73p31c7533e743283f0@mail.gmail.com> "Alex Riesen" writes: > On 1/10/07, Junio C Hamano wrote: >> This comes on top of yours. >> >> I'm reproducing all the merges in linux-2.6 history to make sure >> the base one, yours and this produce the same result (the same >> clean merge, or the same unmerged index and the same diff from >> HEAD). So far it is looking good. > > Yep. Tried the monster merge on it: 1m15sec on that small laptop. Is that supposed to be a good news? It sounds awfully slow. > For whatever reason your patch left an "if (cache_dirty) flush_cache()", > that's after my patch + yours. Had it removed. That's because my copy of "your patch" has the fix-up I suggested to remove the flush from process_renames() already -- the removal of that one and removal from process_entry() you did logically belong to each other.