Re: [PATCH v3] Enable minimal stat checking

From: Johannes Sixt <j.sixt@viscovery.net>
Date: Jan 22, 2013, 08:25 UTC
Message-ID: <50FE4CEB.5040200@viscovery.net>
In-Reply-To: <1358840962-12316-1-git-send-email-robin.rosenberg@dewire.com>

Am 1/22/2013 8:49, schrieb Robin Rosenberg:

Show 9 quoted lines

> Specifically the fields uid, gid, ctime, ino and dev are set to zero
> by JGit. Other implementations, eg. Git in cygwin are allegedly also
> somewhat incompatible with Git For Windows and on *nix platforms
> the resolution of the timestamps may differ.
> 
> Any stat checking by git will then need to check content, which may
> be very slow, particularly on Windows. Since mtime and size
> is typically enough we should allow the user to tell git to avoid
> checking these fields if they are set to zero in the index.

Isn't this paragraph about slowness in the commit message misleading, as what the patch does has no influence on the speed of stat checking? Am I missing something?

> This change introduces a core.checkstat config option where the
> the user can select to check all fields (default), or just size
> and the whole second part of mtime (minimal).

Show 5 quoted lines

> +core.checkstat::
> +	Determines which stat fields to match between the index
> +	and work tree. The user can set this to 'default' or
> +	'minimal'. Default (or explicitly 'default'), is to check
> +	all fields, including the sub-second part of mtime and ctime.

I think this needs some more clarification, less 1337 speak, as well as a hint when to set the option.

	Determines which file attributes are checked to detect whether
	a file has been modified. Set this option to 'minimal', when...,
	which checks only the file size and whole-seconds of the last
	modification time. Otherwise, leave unset or set to the value
	'default'.

By starting with the hint when to set to 'minimal' in this way allows us to omit a specification what the 'default' is.

Show 13 quoted lines

> diff --git a/read-cache.c b/read-cache.c
> index fda78bc..23db681 100644
> --- a/read-cache.c
> +++ b/read-cache.c
> @@ -197,21 +197,25 @@ static int ce_match_stat_basic(struct cache_entry *ce, struct stat *st)
>  	}
>  	if (ce->ce_mtime.sec != (unsigned int)st->st_mtime)
>  		changed |= MTIME_CHANGED;
> -	if (trust_ctime && ce->ce_ctime.sec != (unsigned int)st->st_ctime)
> -		changed |= CTIME_CHANGED;
> +	if (trust_ctime ? check_stat : trust_ctime/*false*/)
> +		if (ce->ce_ctime.sec != (unsigned int)st->st_ctime)
> +			changed |= CTIME_CHANGED;

It took me a while to understand why you write /*false*/ there. Isn't the the condition merely this:

	if (trust_ctime && check_stat &&
	    ce->ce_ctime.sec != (unsigned int)st->st_ctime)
		changed |= CTIME_CHANGED;

Show 10 quoted lines

>  
>  #ifdef USE_NSEC
> -	if (ce->ce_mtime.nsec != ST_MTIME_NSEC(*st))
> +	if (check_stat && ce->ce_mtime.nsec != ST_MTIME_NSEC(*st))
>  		changed |= MTIME_CHANGED;
> -	if (trust_ctime && ce->ce_ctime.nsec != ST_CTIME_NSEC(*st))
> -		changed |= CTIME_CHANGED;
> +	if (trust_ctime ? check_stat : trust_ctime/*false*/)
> +		if (ce->ce_ctime.nsec != ST_CTIME_NSEC(*st))
> +			changed |= CTIME_CHANGED;

Same here.

>  #endif

-- Hannes