All of lore.kernel.org
 help / color / mirror / Atom feed
From: mhagger@alum.mit.edu
To: Junio C Hamano <gitster@pobox.com>
Cc: git@vger.kernel.org, Jeff King <peff@peff.net>,
	Jakub Narebski <jnareb@gmail.com>,
	Heiko Voigt <hvoigt@hvoigt.net>,
	Johan Herland <johan@herland.net>,
	Christian Couder <chriscool@tuxfamily.org>,
	Michael Haggerty <mhagger@alum.mit.edu>
Subject: [PATCH v2 00/18] Read loose references lazily
Date: Fri, 27 Apr 2012 00:26:49 +0200	[thread overview]
Message-ID: <1335479227-7877-1-git-send-email-mhagger@alum.mit.edu> (raw)

From: Michael Haggerty <mhagger@alum.mit.edu>

This is the next installment of the ref-api saga.  The main result of
this patch series is to teach git to read loose references one
directory at a time, only when they are needed.

This version of the patch series differs from v1 by avoiding the
conversion of code from using (struct ref_dir *) to (struct ref_entry
*).  As Junio rightly pointed out, the use of ref_entry as the data
type for dealing with directory entries was awkward and error-prone.

This version also changes how bits in ref_entry::flag are assigned,
also at Junio's suggestion.

The first nine patches are unchanged from v1 except for a trivial
change to a commit message.  These patches convert get_ref_dir() and
do_for_each_reflog() to use strbufs and to tighten up their
specifications.

Patch 10 fixes a git_path() related bug in bisect.c.  A filename
obtained from git_path() was being used long after it was created.
With the changes later in this patch series, the
lazy-loose-reference-reading code will be activated while
check_good_are_ancestors_of_bad() is still holding on to the filename.
Therefore, this fix is a prerequisite to the later patches.

[I am nervous that there might be analogous bugs elsewhere in the
code.  Since the lazy-loose-reference-reading code can now spring to
life in many situations where references are used, it might flush out
problems in sloppy code elsewhere.  But what can we do, short of
getting rid of get_pathname() once and for all?]

Patch 11 Converts find_containing_dir() to use a strbuf.  It makes the
code a bit more straightforward but is otherwise not very interesting.

The rest of the patches teach the code to read loose references
lazily.  This is done by setting a REF_INCOMPLETE bit in
ref_entry::flag for loose directory entries that haven't been read
yet.  The entries are read when they are accessed, namely when
get_ref_dir() is called to extract a ref_dir from a directory
ref_entry.

This patch series depends on mh/ref-api, which just recently made it
to master.  It passes all tests.

Michael Haggerty (18):
  get_ref_dir(): return early if directory cannot be read
  get_ref_dir(): use a strbuf to hold refname
  get_ref_dir(): rename "base" parameter to "dirname"
  get_ref_dir(): require that the dirname argument ends in '/'
  refs.c: extract function search_for_subdir()
  get_ref_dir(): take the containing directory as argument
  do_for_each_reflog(): return early on error
  do_for_each_reflog(): use a strbuf to hold logfile name
  do_for_each_reflog(): reuse strbuf across recursive function calls
  bisect: copy filename string obtained from git_path()
  find_containing_dir(): use strbuf in implementation of this function
  refs: wrap top-level ref_dirs in ref_entries
  read_loose_refs(): rename function from get_ref_dir()
  get_ref_dir(): add function for getting a ref_dir from a ref_entry
  search_for_subdir(): return (ref_dir *) instead of (ref_entry *)
  struct ref_dir: store a reference to the enclosing ref_cache
  read_loose_refs(): eliminate ref_cache argument
  refs: read loose references lazily

 bisect.c |    8 +-
 refs.c   |  378 ++++++++++++++++++++++++++++++++++++++++----------------------
 2 files changed, 247 insertions(+), 139 deletions(-)

-- 
1.7.10

             reply	other threads:[~2012-04-26 22:27 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-04-26 22:26 mhagger [this message]
2012-04-26 22:26 ` [PATCH v2 01/18] get_ref_dir(): return early if directory cannot be read mhagger
2012-04-26 22:26 ` [PATCH v2 02/18] get_ref_dir(): use a strbuf to hold refname mhagger
2012-04-26 22:26 ` [PATCH v2 03/18] get_ref_dir(): rename "base" parameter to "dirname" mhagger
2012-04-26 22:26 ` [PATCH v2 04/18] get_ref_dir(): require that the dirname argument ends in '/' mhagger
2012-04-26 22:26 ` [PATCH v2 05/18] refs.c: extract function search_for_subdir() mhagger
2012-05-03 19:48   ` Junio C Hamano
2012-05-03 20:56     ` Junio C Hamano
2012-05-04  7:24       ` Michael Haggerty
2012-04-26 22:26 ` [PATCH v2 06/18] get_ref_dir(): take the containing directory as argument mhagger
2012-04-26 22:26 ` [PATCH v2 07/18] do_for_each_reflog(): return early on error mhagger
2012-04-26 22:26 ` [PATCH v2 08/18] do_for_each_reflog(): use a strbuf to hold logfile name mhagger
2012-04-26 23:25   ` Junio C Hamano
2012-04-27  8:59     ` Michael Haggerty
2012-05-02 20:06       ` Junio C Hamano
2012-05-03  6:47         ` Michael Haggerty
2012-04-26 22:26 ` [PATCH v2 09/18] do_for_each_reflog(): reuse strbuf across recursive function calls mhagger
2012-04-26 22:26 ` [PATCH v2 10/18] bisect: copy filename string obtained from git_path() mhagger
2012-04-26 22:27 ` [PATCH v2 11/18] find_containing_dir(): use strbuf in implementation of this function mhagger
2012-04-26 22:27 ` [PATCH v2 12/18] refs: wrap top-level ref_dirs in ref_entries mhagger
2012-04-26 22:27 ` [PATCH v2 13/18] read_loose_refs(): rename function from get_ref_dir() mhagger
2012-04-26 22:27 ` [PATCH v2 14/18] get_ref_dir(): add function for getting a ref_dir from a ref_entry mhagger
2012-04-26 22:27 ` [PATCH v2 15/18] search_for_subdir(): return (ref_dir *) instead of (ref_entry *) mhagger
2012-04-26 22:27 ` [PATCH v2 16/18] struct ref_dir: store a reference to the enclosing ref_cache mhagger
2012-04-26 22:27 ` [PATCH v2 17/18] read_loose_refs(): eliminate ref_cache argument mhagger
2012-04-26 22:27 ` [PATCH v2 18/18] refs: read loose references lazily mhagger

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=1335479227-7877-1-git-send-email-mhagger@alum.mit.edu \
    --to=mhagger@alum.mit.edu \
    --cc=chriscool@tuxfamily.org \
    --cc=git@vger.kernel.org \
    --cc=gitster@pobox.com \
    --cc=hvoigt@hvoigt.net \
    --cc=jnareb@gmail.com \
    --cc=johan@herland.net \
    --cc=peff@peff.net \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.