git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Thomas Rast <trast@inf.ethz.ch>
To: Thomas Gummerer <t.gummerer@gmail.com>
Cc: <git@vger.kernel.org>, <gitster@pobox.com>,
	<mhagger@alum.mit.edu>, <pclouds@gmail.com>
Subject: Re: [PATCH/RFC] Replace ce_namelen() with a ce_namelen field
Date: Wed, 4 Jul 2012 12:00:48 +0200	[thread overview]
Message-ID: <87txxn3kxb.fsf@thomas.inf.ethz.ch> (raw)
In-Reply-To: <1341393528-21037-1-git-send-email-t.gummerer@gmail.com> (Thomas Gummerer's message of "Wed, 4 Jul 2012 11:18:48 +0200")

Thomas Gummerer <t.gummerer@gmail.com> writes:

> Replace the ce_namelen() function in cache.h with a ce_namelen
> field in struct cache_entry. This will both give us a tiny bit
> of a performance enhancement when working with long pathnames
> and is part of the refactoring for the index-v5 file format.

Expand this, at least proportionally to the damage incurred.

For example, I know that you know that there's a good reason why we want
to put it in a new field when reading from v5 :-)

> Signed-off-by: Thomas Gummerer <t.gummerer@gmail.com>
> ---
>  builtin/add.c            |    4 ++--
>  builtin/apply.c          |    1 +
>  builtin/blame.c          |    1 +
>  builtin/checkout-index.c |    8 ++++----
>  builtin/checkout.c       |    7 ++++---
>  builtin/clean.c          |    2 +-
>  builtin/commit.c         |    2 +-
>  builtin/grep.c           |    2 +-
>  builtin/ls-files.c       |    8 ++++----
>  builtin/rm.c             |    2 +-
>  builtin/update-index.c   |    9 ++++++---
>  cache-tree.c             |    4 ++--
>  cache.h                  |   11 ++---------
>  diff-lib.c               |    6 +++---
>  entry.c                  |    2 +-
>  merge-recursive.c        |    2 +-
>  name-hash.c              |    4 ++--
>  preload-index.c          |    2 +-
>  read-cache.c             |   43 ++++++++++++++++++++++++-------------------
>  rerere.c                 |    2 +-
>  resolve-undo.c           |    2 +-
>  sha1_name.c              |    6 +++---
>  submodule.c              |    2 +-
>  tree.c                   |    1 +
>  unpack-trees.c           |   31 ++++++++++++++++---------------
>  25 files changed, 85 insertions(+), 79 deletions(-)
[...]
> diff --git a/cache.h b/cache.h
> index cc5048c..5f93f22 100644
> --- a/cache.h
> +++ b/cache.h
> @@ -128,6 +128,7 @@ struct cache_entry {
>  	unsigned int ce_gid;
>  	unsigned int ce_size;
>  	unsigned int ce_flags;
> +	unsigned int ce_namelen;
>  	unsigned char sha1[20];
>  	struct cache_entry *next;
>  	struct cache_entry *dir_next;
> @@ -205,15 +206,7 @@ static inline unsigned create_ce_flags(size_t len, unsigned stage)
>  	return (len | (stage << CE_STAGESHIFT));
>  }
>  
> -static inline size_t ce_namelen(const struct cache_entry *ce)
> -{
> -	size_t len = ce->ce_flags & CE_NAMEMASK;
> -	if (len < CE_NAMEMASK)
> -		return len;
> -	return strlen(ce->name + CE_NAMEMASK) + CE_NAMEMASK;
> -}

AFAICT almost all of the changes relate to the change from ce_namelen(ce)
to ce->ce_namelen.  What's stopping you from defining

#define ce_namelen(ce) ((ce)->ce_namelen)

as a simple compatibility macro?

I'm not sure whether such internal backwards-compatibility is desirable,
but it would certainly make the patch much more reviewable.  If Junio
would rather see the conversion, maybe do it in a separate patch that
simply substitutes the ce->ce_namelen for all the ce_namelen(ce)?

-- 
Thomas Rast
trast@{inf,student}.ethz.ch

  reply	other threads:[~2012-07-04 10:00 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-07-04  9:18 [PATCH/RFC] Replace ce_namelen() with a ce_namelen field Thomas Gummerer
2012-07-04 10:00 ` Thomas Rast [this message]
2012-07-04 10:01 ` Nguyen Thai Ngoc Duy
2012-07-04 10:07 ` Nguyen Thai Ngoc Duy
2012-07-04 19:26 ` Junio C Hamano
2012-07-06 16:07   ` Introduction of " Thomas Gummerer
2012-07-06 16:07     ` [PATCH/RFC v2 1/2] Strip namelen out of ce_flags into " Thomas Gummerer
2012-07-09  4:59       ` Junio C Hamano
2012-07-11  9:22         ` [PATCH v3 0/3] Introduction of " Thomas Gummerer
2012-07-11  9:22           ` [PATCH v3 1/3] read-cache.c: Handle long filenames correctly Thomas Gummerer
2012-07-11 11:34             ` Nguyen Thai Ngoc Duy
2012-07-11 16:33             ` Junio C Hamano
2012-07-11  9:22           ` [PATCH v3 2/3] Strip namelen out of ce_flags into a ce_namelen field Thomas Gummerer
2012-07-11  9:22           ` [PATCH v3 3/3] Replace strlen() with ce_namelen() Thomas Gummerer
2012-07-11 16:35             ` Junio C Hamano
2012-07-11 16:29         ` [PATCH/RFC v2 1/2] Strip namelen out of ce_flags into a ce_namelen field Junio C Hamano
2012-07-06 16:07     ` [PATCH/RFC v2 2/2] Replace strlen() with ce_namelen() Thomas Gummerer
2012-07-09  2:49       ` Junio C Hamano

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=87txxn3kxb.fsf@thomas.inf.ethz.ch \
    --to=trast@inf.ethz.ch \
    --cc=git@vger.kernel.org \
    --cc=gitster@pobox.com \
    --cc=mhagger@alum.mit.edu \
    --cc=pclouds@gmail.com \
    --cc=t.gummerer@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).