From: Christian Couder <christian.couder@gmail.com>
To: Stephan Beyer <s-beyer@gmx.net>
Cc: git <git@vger.kernel.org>, Junio C Hamano <gitster@pobox.com>
Subject: Re: [PATCH v2 19/21] bisect: use a bottom-up traversal to find relevant weights
Date: Wed, 13 Apr 2016 16:11:01 +0200 [thread overview]
Message-ID: <CAP8UFD2_3jwGVJaoOzrKESi8qBN=4D49kiZes+bJARxMvAi3tg@mail.gmail.com> (raw)
In-Reply-To: <1460294354-7031-20-git-send-email-s-beyer@gmx.net>
On Sun, Apr 10, 2016 at 3:19 PM, Stephan Beyer <s-beyer@gmx.net> wrote:
> The idea is to reverse the DAG and perform a traversal
> starting on all sources of the reversed DAG.
>
> We walk from the bottom commits, incrementing the weight while
> walking on a part of the graph that is single strand of pearls,
> or doing the "count the reachable ones the hard way" using
> compute_weight() when we hit a merge commit.
>
> A traversal ends when the computed weight is falling or halfway.
Yeah, it looks like it could be a good optimization to end a traversal
looking for "relevant" commits when the weight is falling.
> This way, commits with too high weight to be relevant are never
> visited (and their weights are never computed).
>
> Signed-off-by: Stephan Beyer <s-beyer@gmx.net>
> ---
>
> Notes:
> I rephrased the commit message.
>
> I renamed the functions such that they don't talk about "BFS"
> because that is irrelevant. Also use a DFS now because it is
> less code (and a little more efficient).
>
> I plugged some leaks.
That's a lot of things in just one commit.
> bisect.c | 250 +++++++++++++++++++++++++++++++++++++++++----------------------
> 1 file changed, 162 insertions(+), 88 deletions(-)
Also from the diff stats it looks like you add a lot of code in this
commit and the previous one.
I wonder why you are saying that a DFS is less code above then.
The previous patch (18/21) has the following diff stat:
> bisect.c | 116 ++++++++++++++++++++++++++++++++++++++++++++++++++++-----------
> 1 file changed, 97 insertions(+), 19 deletions(-)
And the subsequent patches don't reduce code size overall.
Diff stat for 20/21 is:
> bisect.c | 44 +++++++++++++++++++-------------------------
> 1 file changed, 19 insertions(+), 25 deletions(-)
And diff stat for 21/21 is:
> bisect.c | 18 +++++++++++++-----
> 1 file changed, 13 insertions(+), 5 deletions(-)
So after your patches from 18/21 to 21/21 there are around 150 more
lines of code.
Maybe this is worth it, but I wonder if at least some optimizations,
like for example ending a traversal looking for "relevant" commits
when the weight is falling, could be implemented without changing the
code so much and adding so many lines.
next prev parent reply other threads:[~2016-04-13 14:11 UTC|newest]
Thread overview: 56+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-04-10 13:18 [PATCH v2 00/21] git bisect improvements Stephan Beyer
2016-04-10 13:18 ` [PATCH v2 01/21] bisect: write about `bisect next` in documentation Stephan Beyer
2016-04-10 13:18 ` [PATCH v2 02/21] bisect: allow 'bisect run' if no good commit is known Stephan Beyer
2016-04-10 13:18 ` [PATCH v2 03/21] t/test-lib-functions.sh: generalize test_cmp_rev Stephan Beyer
2016-04-11 0:07 ` Eric Sunshine
2016-04-15 20:00 ` Junio C Hamano
2016-04-24 19:51 ` Stephan Beyer
2016-04-25 18:08 ` Junio C Hamano
2016-04-10 13:18 ` [PATCH v2 04/21] t: use test_cmp_rev() where appropriate Stephan Beyer
2016-04-11 0:07 ` Eric Sunshine
2016-04-15 20:48 ` Junio C Hamano
2016-04-10 13:18 ` [PATCH v2 05/21] t6030: generalize test to not rely on current implementation Stephan Beyer
2016-04-10 13:47 ` Torsten Bögershausen
2016-04-10 19:16 ` Junio C Hamano
2016-04-10 19:37 ` Stephan Beyer
2016-04-11 0:23 ` Eric Sunshine
2016-04-15 21:07 ` Junio C Hamano
2016-04-10 13:18 ` [PATCH v2 06/21] bisect: add test for the bisect algorithm Stephan Beyer
2016-04-15 21:13 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 07/21] bisect: plug the biggest memory leak Stephan Beyer
2016-04-15 21:18 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 08/21] bisect: make bisect compile if DEBUG_BISECT is set Stephan Beyer
2016-04-15 21:22 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 09/21] bisect: make algorithm behavior independent of DEBUG_BISECT Stephan Beyer
2016-04-15 21:25 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 10/21] bisect: get rid of recursion in count_distance() Stephan Beyer
2016-04-15 21:31 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 11/21] bisect: use struct node_data array instead of int array Stephan Beyer
2016-04-12 23:02 ` Christian Couder
2016-04-15 21:47 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 12/21] bisect: replace clear_distance() by unique markers Stephan Beyer
2016-04-12 23:20 ` Christian Couder
2016-04-15 22:07 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 13/21] bisect: use commit instead of commit list as arguments when appropriate Stephan Beyer
2016-04-15 22:08 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 14/21] bisect: extract get_distance() function from code duplication Stephan Beyer
2016-04-15 22:08 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 15/21] bisect: introduce distance_direction() Stephan Beyer
2016-04-15 22:10 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 16/21] bisect: make total number of commits global Stephan Beyer
2016-04-13 13:23 ` Christian Couder
2016-04-15 22:11 ` Junio C Hamano
2016-04-16 0:44 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 17/21] bisect: rename count_distance() to compute_weight() Stephan Beyer
2016-04-13 13:32 ` Christian Couder
2016-04-15 22:12 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 18/21] bisect: prepare for different algorithms based on find_all Stephan Beyer
2016-04-15 22:36 ` Junio C Hamano
2016-04-10 13:19 ` [PATCH v2 19/21] bisect: use a bottom-up traversal to find relevant weights Stephan Beyer
2016-04-13 14:11 ` Christian Couder [this message]
2016-04-15 22:47 ` Junio C Hamano
2016-04-15 22:49 ` Junio C Hamano
2016-04-26 18:27 ` Junio C Hamano
2016-04-10 13:24 ` [PATCH v2 20/21] bisect: compute best bisection in compute_relevant_weights() Stephan Beyer
2016-04-10 13:24 ` [PATCH v2 21/21] bisect: get back halfway shortcut Stephan Beyer
2016-04-15 22:53 ` 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
List information: http://vger.kernel.org/majordomo-info.html
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAP8UFD2_3jwGVJaoOzrKESi8qBN=4D49kiZes+bJARxMvAi3tg@mail.gmail.com' \
--to=christian.couder@gmail.com \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=s-beyer@gmx.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.
Code repositories for project(s) associated with this public inbox
https://80x24.org/mirrors/git.git
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 read-only IMAP folder(s) and NNTP newsgroup(s).