From: Duy Nguyen <pclouds@gmail.com>
Date: Tue, 06 May 2014 00:26:26 GMT
Subject: Re: Watchman support for git
Message-ID: <CACsJy8DWK9ysCJobh4oQZe2hxdTaE=n7hH=Nm6bfWzx+yJ=KZg@mail.gmail.com>
In-Reply-To: <CACsJy8C72QQZd4v+p4bkFFKHnN2Uj1zD-BELwGtYB7sx1P6q6g@mail.gmail.com>

On Sat, May 3, 2014 at 7:52 AM, Duy Nguyen <pclouds@gmail.com> wrote:
> wt_status_collect_changes_index() depends on how damaged cache-tree is
> (in this case, totally scraped). watchman does not help this either.
> We need to try to "heal" cache-tree as much as possible to reduce the
> number.

On the topic of cache-tree, I notice that unpack_trees() will call
discard_index() in the end (or even discard_index() on a merge
failure). That destroys cache-tree. unpack_trees() is the core of
branch switching, or reset/merge, which I consider frequent
operations. Cache-tree destruction is bad for "git diff --cached",
"git commit" and maybe more. This discard_index() code was added
recently in e28f764 (unpack-trees: plug a memory leak - 2013-08-13).
As a workaround, perhaps we only do so when the sequencer is running.
-- 
Duy