Re: two questions about the format of loose object

From: Nicolas Pitre <nico@cam.org>
Date: Dec 4, 2008, 00:54 UTC
Message-ID: <alpine.LFD.2.00.0812031949060.14328@xanadu.home>
In-Reply-To: <4934A5EC.2090708@gmail.com>

On Tue, 2 Dec 2008, Liu Yubao wrote:

Show 9 quoted lines

> In fact the format I proposed in my patches is uncompressed loose
> object, not uncompressed loose object header, that's to say I
> proposed format 2 in my question 2, I am just curious why the
> loose object header is compressed in question 1.
> 
> I did a test to add all files of git-1.6.1-rc1 with git-add, the
> time spent decreased by half. Other commands like git diff,
> git diff --cached, git diff HEAD~ HEAD should be faster now
> although the change may be not noticable for small and medium project.

Please try this with an unmodified git version:

	git config --global core.loosecompression 0

and redo your tests please.

One thing that a purely uncompressed loose object format is missing is quick data integrity protection. With the above, you'll have all your loose objects uncompressed but they'll still have a CRC32 done over them.

Nicolas