In this patch, we use the bitmap API to perform the `Counting Objects`
phase in pack-objects, rather than a traditional walk through the object
graph. For a reasonably-packed large repo, the time to fetch and clone
is often dominated by the full-object revision walk during the Counting
Objects phase. Using bitmaps can reduce the CPU time required on the
server (and therefore start sending the actual pack data with less
delay).
For bitmaps to be used, the following must be true:
1. We must be packing to stdout (as a normal `pack-objects` from
`upload-pack` would do).
2. There must be a .bitmap index containing at least one of the
"have" objects that the client is asking for.
3. Bitmaps must be enabled (they are enabled by default, but can be
disabled by setting `pack.usebitmaps` to false, or by using
`--no-use-bitmap-index` on the command-line).
If any of these is not true, we fall back to doing a normal walk of the
object graph.
Here are some sample timings from a full pack of `torvalds/linux` (i.e.
something very similar to what would be generated for a clone of the
repository) that show the speedup produced by various
methods:
[existing graph traversal]
$ time git pack-objects --all --stdout --no-use-bitmap-index \
</dev/null >/dev/null
Counting objects: 3237103, done.
Compressing objects: 100% (508752/508752), done.
Total 3237103 (delta 2699584), reused 3237103 (delta 2699584)
real 0m44.111s
user 0m42.396s
sys 0m3.544s
[bitmaps only, without partial pack reuse; note that
pack reuse is automatic, so timing this required a
patch to disable it]
$ time git pack-objects --all --stdout </dev/null >/dev/null
Counting objects: 3237103, done.
Compressing objects: 100% (508752/508752), done.
Total 3237103 (delta 2699584), reused 3237103 (delta 2699584)
real 0m5.413s
user 0m5.604s
sys 0m1.804s
[bitmaps with pack reuse (what you get with this patch)]
$ time git pack-objects --all --stdout </dev/null >/dev/null
Reusing existing pack: 3237103, done.
Total 3237103 (delta 0), reused 0 (delta 0)
real 0m1.636s
user 0m1.460s
sys 0m0.172s
Signed-off-by: Vicent Marti <tanoku@gmail.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This is the technical documentation for the JGit-compatible Bitmap v1
on-disk format.
Signed-off-by: Vicent Marti <tanoku@gmail.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Two packfiles that contain the same set of objects have
traditionally been named identically, but that made repacking a
repository that is already fully packed without any cruft with a
different packing parameter cumbersome. Update the convention to
name the packfile after the bytestream representation of the data,
not after the set of objects in it.
* jk/name-pack-after-byte-representation:
pack-objects doc: treat output filename as opaque
pack-objects: name pack files after trailer hash
sha1write: make buffer const-correct
Show the total number of paths and the number of paths shown so far
when "git difftool" prompts to launch an external diff tool, which
would give users some sense of progress.
* zk/difftool-counts:
diff.c: fix some recent whitespace style violations
difftool: display the number of files in the diff queue in the prompt
Make "git push origin master" update the same ref that would be
updated by our 'master' when "git push origin" (no refspecs) is run
while the 'master' branch is checked out, which makes "git push"
more symmetric to "git fetch" and more usable for the triangular
workflow.
* jc/push-refmap:
push: also use "upstream" mapping when pushing a single ref
push: use remote.$name.push as a refmap
builtin/push.c: use strbuf instead of manual allocation
It can be useful for debugging or analysis to see which
objects are stored as delta bases on top of others. This
information is available by running `git verify-pack`, but
that is extremely expensive (and is harder than necessary to
parse).
Instead, let's make it available as a cat-file query format,
which makes it fast and simple to get the bases for a subset
of the objects.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
diff.orderfile acts as a default for the -O command line option.
[sb: split up aw's original patch; rework tests and docs, treat option
as pathname]
Signed-off-by: Anders Waldenborg <anders@0x63.nu>
Signed-off-by: Samuel Bronson <naesten@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The BFG is a tool specifically designed for the task of removing
unwanted data from Git repository history - a common use-case for which
git-filter-branch has been the traditional workhorse.
It's beneficial to let users know that filter-branch has an alternative
here:
* speed : The BFG is 10-50x faster
http://rtyley.github.io/bfg-repo-cleaner/#speed
* complexity of configuration : filter-branch is a very flexible tool,
but demands very careful usage in order to get the desired results
http://rtyley.github.io/bfg-repo-cleaner/#examples
Obviously, filter-branch has it's advantages too - it permits very
complex rewrites, and doesn't require a JVM - but for the common
use-case of deleting unwanted data, it's helpful to users to be aware
that an alternative exists.
The BFG was released under the GPL in February 2013, and has since seen
widespread production use (The Guardian, RedHat, Google, UK Government
Digital Service), been tested against large repos (~300K commits, ~5GB
packfiles) and received significant positive feedback from users:
http://rtyley.github.io/bfg-repo-cleaner/#feedback
Signed-off-by: Roberto Tyley <roberto.tyley@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Allow gitweb to be configured to show refs out of refs/heads/ as if
they were branches.
* kn/gitweb-extra-branch-refs:
gitweb: Denote non-heads, non-remotes branches
gitweb: Add a feature for adding more branch refs
gitweb: Return 1 on validation success instead of passed input
gitweb: Move check-ref-format code into separate function
After 1190a1a (pack-objects: name pack files after trailer hash,
2013-12-05), the SHA-1 used to determine the filename is calculated
differently. Update the documentation to not guarantee anything more
than that the SHA-1 depends on the pack content somehow.
Hopefully this will discourage readers from depending on the old or
the new calculation.
Reported-by: Michael Haggerty <mhagger@alum.mit.edu>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Allow receive-pack to insist on receiving a fat pack from "git
push" clients.
* cn/thin-push-capability:
send-pack: don't send a thin pack to a server which doesn't support it
The "--tags" option to "git fetch" used to be literally a synonym to
a "refs/tags/*:refs/tags/*" refspec, which meant that (1) as an
explicit refspec given from the command line, it silenced the lazy
"git fetch" default that is configured, and (2) also as an explicit
refspec given from the command line, it interacted with "--prune"
to remove any tag that the remote we are fetching from does not
have.
This demotes it to an option; with it, we fetch all tags in
addition to what would be fetched without the option, and it does
not interact with the decision "--prune" makes to see what
remote-tracking refs the local has are missing the remote
counterpart.
* mh/fetch-tags-in-addition-to-normal-refs: (23 commits)
fetch: improve the error messages emitted for conflicting refspecs
handle_duplicate(): mark error message for translation
ref_remote_duplicates(): extract a function handle_duplicate()
ref_remove_duplicates(): simplify loop logic
t5536: new test of refspec conflicts when fetching
ref_remove_duplicates(): avoid redundant bisection
git-fetch.txt: improve description of tag auto-following
fetch-options.txt: simplify ifdef/ifndef/endif usage
fetch, remote: properly convey --no-prune options to subprocesses
builtin/remote.c:update(): use struct argv_array
builtin/remote.c: reorder function definitions
query_refspecs(): move some constants out of the loop
fetch --prune: prune only based on explicit refspecs
fetch --tags: fetch tags *in addition to* other stuff
fetch: only opportunistically update references based on command line
get_expanded_map(): avoid memory leak
get_expanded_map(): add docstring
builtin/fetch.c: reorder function definitions
get_ref_map(): rename local variables
api-remote.txt: correct section "struct refspec"
...
Allow extra-branch-refs feature to tell gitweb to show refs from
additional hierarchies in addition to branches in the list-of-branches
view.
Signed-off-by: Krzesimir Nowak <krzesimir@endocode.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Now that git supports data transfer from or to a shallow clone, these
limitations are not true anymore.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch teaches "prune" to remove shallow roots that are no longer
reachable from any refs (e.g. when the relevant refs are removed).
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The basic 8 steps to update .git/shallow does not fully apply here
because the user may choose to accept just a few refs (while fetch
always accepts all refs). The steps are modified a bit.
1-6. same as before. After calling assign_shallow_commits_to_refs at
step 6, each shallow commit has a bitmap that marks all refs that
require it.
7. mark all "ours" shallow commits that are reachable from any
refs. We will need to do the original step 7 on them later.
8. go over all shallow commit bitmaps, mark refs that require new
shallow commits.
9. setup a strict temporary shallow file to plug all the holes, even
if it may cut some of our history short. This file is used by all
hooks. The hooks could use --shallow-file=$GIT_DIR/shallow to
overcome this and reach everything in current repo.
10. go over the new refs one by one. For each ref, do the reachability
test if it needs a shallow commit on the list from step 7. Remove
it if it's reachable from our refs. Gather all required shallow
commits, run check_everything_connected() with the new ref, then
install them to .git/shallow.
This mode is disabled by default and can be turned on with
receive.shallowupdate
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The same steps are done as in when --update-shallow is not given. The
only difference is we now add all shallow commits in "ours" and
"theirs" to .git/shallow (aka "step 8").
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When "fetch --depth=N" where N exceeds the longest chain of history in
the source repo, usually we just send an "unshallow" line to the
client so full history is obtained.
When the source repo is shallow we need to make sure to "unshallow"
the current shallow point _and_ "shallow" again when the commit
reaches its shallow bottom in the source repo.
This should fix both cases: large <N> and --unshallow.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If either receive-pack or upload-pack is called on a shallow
repository, shallow commits (*) will be sent after the ref
advertisement (but before the packet flush), so that the receiver has
the full "shape" of the sender's commit graph. This will be needed for
the receiver to update its .git/shallow if necessary.
This breaks the protocol for all clients trying to push to a shallow
repo, or fetch from one. Which is basically the same end result as
today's "is_repository_shallow() && die()" in receive-pack and
upload-pack. New clients will be made aware of shallow upstream and
can make use of this information.
The sender must send all shallow commits that are sent in the
following pack. It may send more shallow commits than necessary.
upload-pack for example may choose to advertise no shallow commits if
it knows in advance that the pack it's going to send contains no
shallow commits. But upload-pack is the server, so we choose the
cheaper way, send full .git/shallow and let the client deal with it.
Smart HTTP is not affected by this patch. Shallow support on
smart-http comes later separately.
(*) A shallow commit is a commit that terminates the revision
walker. It is usually put in .git/shallow in order to keep the
revision walker from going out of bound because there is no
guarantee that objects behind this commit is available.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit 15a147e (rebase: use @{upstream} if no upstream specified,
2011-02-09) says:
Make it default to 'git rebase @{upstream}'. That is also what
'git pull [--rebase]' defaults to, so it only makes sense that
'git rebase' defaults to the same thing.
but that isn't actually the case. Since commit d44e712 (pull: support
rebased upstream + fetch + pull --rebase, 2009-07-19), pull has actually
chosen the most recent reflog entry which is an ancestor of the current
branch if it can find one.
Add a '--fork-point' argument to git-rebase that can be used to trigger
this behaviour. This option is turned on by default if no non-option
arguments are specified on the command line, otherwise we treat an
upstream specified on the command-line literally.
Signed-off-by: John Keeping <john@keeping.me.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Oftentimes people will make the same change in two branches, revert the change
in one branch, and then be surprised when a merge reinstitutes that change when
the branches are merged. Add an explanatory paragraph that explains that this
occurs and the reason why, so people are not surprised.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There's no mention of the 'origin' default, or the fact that the
upstream tracking branch remote is used.
Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Additionally, precedence of negated patterns is exactly as outlined in
the DESCRIPTION section, we don't need to repeat this.
Signed-off-by: Karsten Blees <blees@dcon.de>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When --prompt option is set, git-difftool displays a prompt for each
modified file to be viewed in an external diff program. At that
point, it could be useful to display a counter and the total number
of files in the diff queue.
Below is the current difftool prompt for the first of 5 modified files:
Viewing: 'diff.c'
Launch 'vimdiff' [Y/n]:
Consider the modified prompt:
Viewing (1/5): 'diff.c'
Launch 'vimdiff' [Y/n]:
The current GIT_EXTERNAL_DIFF mechanism does not tell the number of
paths in the diff queue nor the current counter. To make this
"counter/total" info available for GIT_EXTERNAL_DIFF programs
without breaking existing ones by doing the following:
- Keep track of the number of paths shown so far in diff_options;
- Export two new environment variables from run_external_diff() to
show the total number of paths (from diff_queue_struct) and the
current value of the counter (from diff_options); and
- Update git-difftool--helper to use these two environment variables.
Signed-off-by: Zoltan Klinger <zoltan.klinger@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
