"git pack-objects" learned to invoke a new hook program that
enumerates extra objects to be used as anchoring points to keep
otherwise unreachable objects in cruft packs.
* tb/gc-recent-object-hook:
gc: introduce `gc.recentObjectsHook`
reachable.c: extract `obj_is_recent()`
The object traversal using reachability bitmap done by
"pack-object" has been tweaked to take advantage of the fact that
using "boundary" commits as representative of all the uninteresting
ones can save quite a lot of object enumeration.
* tb/pack-bitmap-traversal-with-boundary:
pack-bitmap.c: use commit boundary during bitmap traversal
pack-bitmap.c: extract `fill_in_bitmap()`
object: add object_array initializer helper function
'git worktree add' learned how to create a worktree based on an
orphaned branch with `--orphan`.
* ja/worktree-orphan:
worktree add: emit warn when there is a bad HEAD
worktree add: extend DWIM to infer --orphan
worktree add: introduce "try --orphan" hint
worktree add: add --orphan flag
t2400: add tests to verify --quiet
t2400: refactor "worktree add" opt exclusion tests
t2400: cleanup created worktree in test
worktree add: include -B in usage docs
This patch introduces a new multi-valued configuration option,
`gc.recentObjectsHook` as a means to mark certain objects as recent (and
thus exempt from garbage collection), regardless of their age.
When performing a garbage collection operation on a repository with
unreachable objects, Git makes its decision on what to do with those
object(s) based on how recent the objects are or not. Generally speaking,
unreachable-but-recent objects stay in the repository, and older objects
are discarded.
However, we have no convenient way to keep certain precious, unreachable
objects around in the repository, even if they have aged out and would
be pruned. Our options today consist of:
- Point references at the reachability tips of any objects you
consider precious, which may be undesirable or infeasible if there
are many such objects.
- Track them via the reflog, which may be undesirable since the
reflog's lifetime is limited to that of the reference it's tracking
(and callers may want to keep those unreachable objects around for
longer).
- Extend the grace period, which may keep around other objects that
the caller *does* want to discard.
- Manually modify the mtimes of objects you want to keep. If those
objects are already loose, this is easy enough to do (you can just
enumerate and `touch -m` each one).
But if they are packed, you will either end up modifying the mtimes
of *all* objects in that pack, or be forced to write out a loose
copy of that object, both of which may be undesirable. Even worse,
if they are in a cruft pack, that requires modifying its `*.mtimes`
file by hand, since there is no exposed plumbing for this.
- Force the caller to construct the pack of objects they want
to keep themselves, and then mark the pack as kept by adding a
".keep" file. This works, but is burdensome for the caller, and
having extra packs is awkward as you roll forward your cruft pack.
This patch introduces a new option to the above list via the
`gc.recentObjectsHook` configuration, which allows the caller to
specify a program (or set of programs) whose output is treated as a set
of objects to treat as recent, regardless of their true age.
The implementation is straightforward. Git enumerates recent objects via
`add_unseen_recent_objects_to_traversal()`, which enumerates loose and
packed objects, and eventually calls add_recent_object() on any objects
for which `want_recent_object()`'s conditions are met.
This patch modifies the recency condition from simply "is the mtime of
this object more recent than the cutoff?" to "[...] or, is this object
mentioned by at least one `gc.recentObjectsHook`?".
Depending on whether or not we are generating a cruft pack, this allows
the caller to do one of two things:
- If generating a cruft pack, the caller is able to retain additional
objects via the cruft pack, even if they would have otherwise been
pruned due to their age.
- If not generating a cruft pack, the caller is likewise able to
retain additional objects as loose.
A potential alternative here is to introduce a new mode to alter the
contents of the reachable pack instead of the cruft one. One could
imagine a new option to `pack-objects`, say `--extra-reachable-tips`
that does the same thing as above, adding the visited set of objects
along the traversal to the pack.
But this has the unfortunate side-effect of altering the reachability
closure of that pack. If parts of the unreachable object graph mentioned
by one or more of the "extra reachable tips" programs is not closed,
then the resulting pack won't be either. This makes it impossible in the
general case to write out reachability bitmaps for that pack, since
closure is a requirement there.
Instead, keep these unreachable objects in the cruft pack (or set of
unreachable, loose objects) instead, to ensure that we can continue to
have a pack containing just reachable objects, which is always safe to
write a bitmap over.
Helped-by: Jeff King <peff@peff.net>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a new advice/hint in `git worktree add` for when the user
tries to create a new worktree from a reference that doesn't exist.
Current Behavior:
% git init foo
Initialized empty Git repository in /path/to/foo/
% touch file
% git -C foo commit -q -a -m "test commit"
% git -C foo switch --orphan norefbranch
% git -C foo worktree add newbranch/
Preparing worktree (new branch 'newbranch')
fatal: invalid reference: HEAD
%
New Behavior:
% git init --bare foo
Initialized empty Git repository in /path/to/foo/
% touch file
% git -C foo commit -q -a -m "test commit"
% git -C foo switch --orphan norefbranch
% git -C foo worktree add newbranch/
Preparing worktree (new branch 'newbranch')
hint: If you meant to create a worktree containing a new orphan branch
hint: (branch with no commits) for this repository, you can do so
hint: using the --orphan option:
hint:
hint: git worktree add --orphan newbranch/
hint:
hint: Disable this message with "git config advice.worktreeAddOrphan false"
fatal: invalid reference: HEAD
% git -C foo worktree add -b newbranch2 new_wt/
Preparing worktree (new branch 'newbranch')
hint: If you meant to create a worktree containing a new orphan branch
hint: (branch with no commits) for this repository, you can do so
hint: using the --orphan option:
hint:
hint: git worktree add --orphan -b newbranch2 new_wt/
hint:
hint: Disable this message with "git config advice.worktreeAddOrphan false"
fatal: invalid reference: HEAD
%
Signed-off-by: Jacob Abel <jacobabel@nullpo.dev>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add the unit (bytes per second) for http.lowSpeedLimit
in the documentation.
Signed-off-by: Corentin Garcia <corenting@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When reachability bitmap coverage exists in a repository, Git will use a
different (and hopefully faster) traversal to compute revision walks.
Consider a set of positive and negative tips (which we'll refer to with
their standard bitmap parlance by "wants", and "haves"). In order to
figure out what objects exist between the tips, the existing traversal
in `prepare_bitmap_walk()` does something like:
1. Consider if we can even compute the set of objects with bitmaps,
and fall back to the usual traversal if we cannot. For example,
pathspec limiting traversals can't be computed using bitmaps (since
they don't know which objects are at which paths). The same is true
of certain kinds of non-trivial object filters.
2. If we can compute the traversal with bitmaps, partition the
(dereferenced) tips into two object lists, "haves", and "wants",
based on whether or not the objects have the UNINTERESTING flag,
respectively.
3. Fall back to the ordinary object traversal if either (a) there are
more than zero haves, none of which are in the bitmapped pack or
MIDX, or (b) there are no wants.
4. Construct a reachability bitmap for the "haves" side by walking
from the revision tips down to any existing bitmaps, OR-ing in any
bitmaps as they are found.
5. Then do the same for the "wants" side, stopping at any objects that
appear in the "haves" bitmap.
6. Filter the results if any object filter (that can be easily
computed with bitmaps alone) was given, and then return back to the
caller.
When there is good bitmap coverage relative to the traversal tips, this
walk is often significantly faster than an ordinary object traversal
because it can visit far fewer objects.
But in certain cases, it can be significantly *slower* than the usual
object traversal. Why? Because we need to compute complete bitmaps on
either side of the walk. If either one (or both) of the sides require
walking many (or all!) objects before they get to an existing bitmap,
the extra bitmap machinery is mostly or all overhead.
One of the benefits, however, is that even if the walk is slower, bitmap
traversals are guaranteed to provide an *exact* answer. Unlike the
traditional object traversal algorithm, which can over-count the results
by not opening trees for older commits, the bitmap walk builds an exact
reachability bitmap for either side, meaning the results are never
over-counted.
But producing non-exact results is OK for our traversal here (both in
the bitmap case and not), as long as the results are over-counted, not
under.
Relaxing the bitmap traversal to allow it to produce over-counted
results gives us the opportunity to make some significant improvements.
Instead of the above, the new algorithm only has to walk from the
*boundary* down to the nearest bitmap, instead of from each of the
UNINTERESTING tips.
The boundary-based approach still has degenerate cases, but we'll show
in a moment that it is often a significant improvement.
The new algorithm works as follows:
1. Build a (partial) bitmap of the haves side by first OR-ing any
bitmap(s) that already exist for UNINTERESTING commits between the
haves and the boundary.
2. For each commit along the boundary, add it as a fill-in traversal
tip (where the traversal terminates once an existing bitmap is
found), and perform fill-in traversal.
3. Build up a complete bitmap of the wants side as usual, stopping any
time we intersect the (partial) haves side.
4. Return the results.
And is more-or-less equivalent to using the *old* algorithm with this
invocation:
$ git rev-list --objects --use-bitmap-index $WANTS --not \
$(git rev-list --objects --boundary $WANTS --not $HAVES |
perl -lne 'print $1 if /^-(.*)/')
The new result performs significantly better in many cases, particularly
when the distance from the boundary commit(s) to an existing bitmap is
shorter than the distance from (all of) the have tips to the nearest
bitmapped commit.
Note that when using the old bitmap traversal algorithm, the results can
be *slower* than without bitmaps! Under the new algorithm, the result is
computed faster with bitmaps than without (at the cost of over-counting
the true number of objects in a similar fashion as the non-bitmap
traversal):
# (Computing the number of tagged objects not on any branches
# without bitmaps).
$ time git rev-list --count --objects --tags --not --branches
20
real 0m1.388s
user 0m1.092s
sys 0m0.296s
# (Computing the same query using the old bitmap traversal).
$ time git rev-list --count --objects --tags --not --branches --use-bitmap-index
19
real 0m22.709s
user 0m21.628s
sys 0m1.076s
# (this commit)
$ time git.compile rev-list --count --objects --tags --not --branches --use-bitmap-index
19
real 0m1.518s
user 0m1.234s
sys 0m0.284s
The new algorithm is still slower than not using bitmaps at all, but it
is nearly a 15-fold improvement over the existing traversal.
In a more realistic setting (using my local copy of git.git), I can
observe a similar (if more modest) speed-up:
$ argv="--count --objects --branches --not --tags"
hyperfine \
-n 'no bitmaps' "git.compile rev-list $argv" \
-n 'existing traversal' "git.compile rev-list --use-bitmap-index $argv" \
-n 'boundary traversal' "git.compile -c pack.useBitmapBoundaryTraversal=true rev-list --use-bitmap-index $argv"
Benchmark 1: no bitmaps
Time (mean ± σ): 124.6 ms ± 2.1 ms [User: 103.7 ms, System: 20.8 ms]
Range (min … max): 122.6 ms … 133.1 ms 22 runs
Benchmark 2: existing traversal
Time (mean ± σ): 368.6 ms ± 3.0 ms [User: 325.3 ms, System: 43.1 ms]
Range (min … max): 365.1 ms … 374.8 ms 10 runs
Benchmark 3: boundary traversal
Time (mean ± σ): 167.6 ms ± 0.9 ms [User: 139.5 ms, System: 27.9 ms]
Range (min … max): 166.1 ms … 169.2 ms 17 runs
Summary
'no bitmaps' ran
1.34 ± 0.02 times faster than 'boundary traversal'
2.96 ± 0.05 times faster than 'existing traversal'
Here, the new algorithm is also still slower than not using bitmaps, but
represents a more than 2-fold improvement over the existing traversal in
a more modest example.
Since this algorithm was originally written (nearly a year and a half
ago, at the time of writing), the bitmap lookup table shipped, making
the new algorithm's result more competitive. A few other future
directions for improving bitmap traversal times beyond not using bitmaps
at all:
- Decrease the cost to decompress and OR together many bitmaps
together (particularly when enumerating the uninteresting side of
the walk). Here we could explore more efficient bitmap storage
techniques, like Roaring+Run and/or use SIMD instructions to speed
up ORing them together.
- Store pseudo-merge bitmaps, which could allow us to OR together
fewer "summary" bitmaps (which would also help with the above).
Helped-by: Jeff King <peff@peff.net>
Helped-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Sometimes, adding a header different than CC or TO is desirable; for
example, when using Debbugs, it is best to use 'X-Debbugs-Cc' headers
to keep people in CC; this is an example use case enabled by the new
'--header-cmd' option.
The header unfolding logic is extracted to a subroutine so that it can
be reused; a test is added for coverage.
Signed-off-by: Maxim Cournoyer <maxim.cournoyer@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When "gc" needs to retain unreachable objects, packing them into
cruft packs (instead of exploding them into loose object files) has
been offered as a more efficient option for some time. Now the use
of cruft packs has been made the default and no longer considered
an experimental feature.
* tb/enable-cruft-packs-by-default:
repository.h: drop unused `gc_cruft_packs`
builtin/gc.c: make `gc.cruftPacks` enabled by default
t/t9300-fast-import.sh: prepare for `gc --cruft` by default
t/t6500-gc.sh: add additional test cases
t/t6500-gc.sh: refactor cruft pack tests
t/t6501-freshen-objects.sh: prepare for `gc --cruft` by default
t/t5304-prune.sh: prepare for `gc --cruft` by default
builtin/gc.c: ignore cruft packs with `--keep-largest-pack`
builtin/repack.c: fix incorrect reference to '-C'
pack-write.c: plug a leak in stage_tmp_packfiles()
The on-disk reverse index that allows mapping from the pack offset
to the object name for the object stored at the offset has been
enabled by default.
* tb/pack-revindex-on-disk:
t: invert `GIT_TEST_WRITE_REV_INDEX`
config: enable `pack.writeReverseIndex` by default
pack-revindex: introduce `pack.readReverseIndex`
pack-revindex: introduce GIT_TEST_REV_INDEX_DIE_ON_DISK
pack-revindex: make `load_pack_revindex` take a repository
t5325: mark as leak-free
pack-write.c: plug a leak in stage_tmp_packfiles()
Back in 5b92477f89 (builtin/gc.c: conditionally avoid pruning objects
via loose, 2022-05-20), `git gc` learned the `--cruft` option and
`gc.cruftPacks` configuration to opt-in to writing cruft packs when
collecting or pruning unreachable objects.
Cruft packs were introduced with the merge in a50036da1a (Merge branch
'tb/cruft-packs', 2022-06-03). They address the problem of "loose object
explosions", where Git will write out many individual loose objects when
there is a large number of unreachable objects that have not yet aged
past `--prune=<date>`.
Instead of keeping track of those unreachable yet recent objects via
their loose object file's mtime, cruft packs collect all unreachable
objects into a single pack with a corresponding `*.mtimes` file that
acts as a table to store the mtimes of all unreachable objects. This
prevents the need to store unreachable objects as loose as they age out
of the repository, and avoids the problem of loose object explosions.
Beyond avoiding loose object explosions, cruft packs also act as a more
efficient mechanism to store unreachable objects as they age out of a
repository. This is because pairs of similar unreachable objects serve
as delta bases for one another.
In 5b92477f89, the feature was introduced as experimental. Since then,
GitHub has been running these patches in every repository generating
hundreds of millions of cruft packs along the way. The feature is
battle-tested, and avoids many pathological cases such as above. Users
who either run `git gc` manually, or via `git maintenance` can benefit
from having cruft packs.
As such, enable cruft pack generation to take place by default (by
making `gc.cruftPacks` have the default of "true" rather than "false).
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When cruft packs were implemented, we never adjusted the code for `git
gc`'s `--keep-largest-pack` and `gc.bigPackThreshold` to ignore cruft
packs. This option and configuration option share a common
implementation, but including cruft packs is wrong in both cases:
- Running `git gc --keep-largest-pack` in a repository where the
largest pack is the cruft pack itself will make it impossible for
`git gc` to prune objects, since the cruft pack itself is kept.
- The same is true for `gc.bigPackThreshold`, if the size of the cruft
pack exceeds the limit set by the caller.
In the future, it is possible that `gc.bigPackThreshold` could be used
to write a separate cruft pack containing any new unreachable objects
that entered the repository since the last time a cruft pack was
written.
There are some complexities to doing so, mainly around handling
pruning objects that are in an existing cruft pack that is above the
threshold (which would either need to be rewritten, or else delay
pruning). Rewriting a substantially similar cruft pack isn't ideal, but
it is significantly better than the status-quo.
If users have large cruft packs that they don't want to rewrite, they
can mark them as `*.keep` packs. But in general, if a repository has a
cruft pack that is so large it is slowing down GC's, it should probably
be pruned anyway.
In the meantime, ignore cruft packs in the common implementation for
both of these options, and add a pair of tests to prevent any future
regressions here.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Back in e37d0b8730 (builtin/index-pack.c: write reverse indexes,
2021-01-25), Git learned how to read and write a pack's reverse index
from a file instead of in-memory.
A pack's reverse index is a mapping from pack position (that is, the
order that objects appear together in a ".pack") to their position in
lexical order (that is, the order that objects are listed in an ".idx"
file).
Reverse indexes are consulted often during pack-objects, as well as
during auxiliary operations that require mapping between pack offsets,
pack order, and index index.
They are useful in GitHub's infrastructure, where we have seen a
dramatic increase in performance when writing ".rev" files[1]. In
particular:
- an ~80% reduction in the time it takes to serve fetches on a popular
repository, Homebrew/homebrew-core.
- a ~60% reduction in the peak memory usage to serve fetches on that
same repository.
- a collective savings of ~35% in CPU time across all pack-objects
invocations serving fetches across all repositories in a single
datacenter.
Reverse indexes are also beneficial to end-users as well as forges. For
example, the time it takes to generate a pack containing the objects for
the 10 most recent commits in linux.git (representing a typical push) is
significantly faster when on-disk reverse indexes are available:
$ { git rev-parse HEAD && printf '^' && git rev-parse HEAD~10 } >in
$ hyperfine -L v false,true 'git.compile -c pack.readReverseIndex={v} pack-objects --delta-base-offset --revs --stdout <in >/dev/null'
Benchmark 1: git.compile -c pack.readReverseIndex=false pack-objects --delta-base-offset --revs --stdout <in >/dev/null
Time (mean ± σ): 543.0 ms ± 20.3 ms [User: 616.2 ms, System: 58.8 ms]
Range (min … max): 521.0 ms … 577.9 ms 10 runs
Benchmark 2: git.compile -c pack.readReverseIndex=true pack-objects --delta-base-offset --revs --stdout <in >/dev/null
Time (mean ± σ): 245.0 ms ± 11.4 ms [User: 335.6 ms, System: 31.3 ms]
Range (min … max): 226.0 ms … 259.6 ms 13 runs
Summary
'git.compile -c pack.readReverseIndex=true pack-objects --delta-base-offset --revs --stdout <in >/dev/null' ran
2.22 ± 0.13 times faster than 'git.compile -c pack.readReverseIndex=false pack-objects --delta-base-offset --revs --stdout <in >/dev/null'
The same is true of writing a pack containing the objects for the 30
most-recent commits:
$ { git rev-parse HEAD && printf '^' && git rev-parse HEAD~30 } >in
$ hyperfine -L v false,true 'git.compile -c pack.readReverseIndex={v} pack-objects --delta-base-offset --revs --stdout <in >/dev/null'
Benchmark 1: git.compile -c pack.readReverseIndex=false pack-objects --delta-base-offset --revs --stdout <in >/dev/null
Time (mean ± σ): 866.5 ms ± 16.2 ms [User: 1414.5 ms, System: 97.0 ms]
Range (min … max): 839.3 ms … 886.9 ms 10 runs
Benchmark 2: git.compile -c pack.readReverseIndex=true pack-objects --delta-base-offset --revs --stdout <in >/dev/null
Time (mean ± σ): 581.6 ms ± 10.2 ms [User: 1181.7 ms, System: 62.6 ms]
Range (min … max): 567.5 ms … 599.3 ms 10 runs
Summary
'git.compile -c pack.readReverseIndex=true pack-objects --delta-base-offset --revs --stdout <in >/dev/null' ran
1.49 ± 0.04 times faster than 'git.compile -c pack.readReverseIndex=false pack-objects --delta-base-offset --revs --stdout <in >/dev/null'
...and savings on trivial operations like computing the on-disk size of
a single (packed) object are even more dramatic:
$ git rev-parse HEAD >in
$ hyperfine -L v false,true 'git.compile -c pack.readReverseIndex={v} cat-file --batch-check="%(objectsize:disk)" <in'
Benchmark 1: git.compile -c pack.readReverseIndex=false cat-file --batch-check="%(objectsize:disk)" <in
Time (mean ± σ): 305.8 ms ± 11.4 ms [User: 264.2 ms, System: 41.4 ms]
Range (min … max): 290.3 ms … 331.1 ms 10 runs
Benchmark 2: git.compile -c pack.readReverseIndex=true cat-file --batch-check="%(objectsize:disk)" <in
Time (mean ± σ): 4.0 ms ± 0.3 ms [User: 1.7 ms, System: 2.3 ms]
Range (min … max): 1.6 ms … 4.6 ms 1155 runs
Summary
'git.compile -c pack.readReverseIndex=true cat-file --batch-check="%(objectsize:disk)" <in' ran
76.96 ± 6.25 times faster than 'git.compile -c pack.readReverseIndex=false cat-file --batch-check="%(objectsize:disk)" <in'
In the more than two years since e37d0b8730 was merged, Git's
implementation of on-disk reverse indexes has been thoroughly tested,
both from users enabling `pack.writeReverseIndexes`, and from GitHub's
deployment of the feature. The latter has been running without incident
for more than two years.
This patch changes Git's behavior to write on-disk reverse indexes by
default when indexing a pack, which should make the above operations
faster for everybody's Git installation after a repack.
(The previous commit explains some potential drawbacks of using on-disk
reverse indexes in certain limited circumstances, that essentially boil
down to a trade-off between time to generate, and time to access. For
those limited cases, the `pack.readReverseIndex` escape hatch can be
used).
[1]: https://github.blog/2021-04-29-scaling-monorepo-maintenance/#reverse-indexes
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Acked-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since 1615c567b8 (Documentation/config/pack.txt: advertise
'pack.writeReverseIndex', 2021-01-25), we have had the
`pack.writeReverseIndex` configuration option, which tells Git whether
or not it is allowed to write a ".rev" file when indexing a pack.
Introduce a complementary configuration knob, `pack.readReverseIndex` to
control whether or not Git will read any ".rev" file(s) that may be
available on disk.
This option is useful for debugging, as well as disabling the effect of
".rev" files in certain instances.
This is useful because of the trade-off[^1] between the time it takes to
generate a reverse index (slow from scratch, fast when reading an
existing ".rev" file), and the time it takes to access a record (the
opposite).
For example, even though it is faster to use the on-disk reverse index
when computing the on-disk size of a packed object, it is slower to
enumerate the same value for all objects.
Here are a couple of examples from linux.git. When computing the above
for a single object, using the on-disk reverse index is significantly
faster:
$ git rev-parse HEAD >in
$ hyperfine -L v false,true 'git.compile -c pack.readReverseIndex={v} cat-file --batch-check="%(objectsize:disk)" <in'
Benchmark 1: git.compile -c pack.readReverseIndex=false cat-file --batch-check="%(objectsize:disk)" <in
Time (mean ± σ): 302.5 ms ± 12.5 ms [User: 258.7 ms, System: 43.6 ms]
Range (min … max): 291.1 ms … 328.1 ms 10 runs
Benchmark 2: git.compile -c pack.readReverseIndex=true cat-file --batch-check="%(objectsize:disk)" <in
Time (mean ± σ): 3.9 ms ± 0.3 ms [User: 1.6 ms, System: 2.4 ms]
Range (min … max): 2.0 ms … 4.4 ms 801 runs
Summary
'git.compile -c pack.readReverseIndex=true cat-file --batch-check="%(objectsize:disk)" <in' ran
77.29 ± 7.14 times faster than 'git.compile -c pack.readReverseIndex=false cat-file --batch-check="%(objectsize:disk)" <in'
, but when instead trying to compute the on-disk object size for all
objects in the repository, using the ".rev" file is a disadvantage over
creating the reverse index from scratch:
$ hyperfine -L v false,true 'git.compile -c pack.readReverseIndex={v} cat-file --batch-check="%(objectsize:disk)" --batch-all-objects'
Benchmark 1: git.compile -c pack.readReverseIndex=false cat-file --batch-check="%(objectsize:disk)" --batch-all-objects
Time (mean ± σ): 8.258 s ± 0.035 s [User: 7.949 s, System: 0.308 s]
Range (min … max): 8.199 s … 8.293 s 10 runs
Benchmark 2: git.compile -c pack.readReverseIndex=true cat-file --batch-check="%(objectsize:disk)" --batch-all-objects
Time (mean ± σ): 16.976 s ± 0.107 s [User: 16.706 s, System: 0.268 s]
Range (min … max): 16.839 s … 17.105 s 10 runs
Summary
'git.compile -c pack.readReverseIndex=false cat-file --batch-check="%(objectsize:disk)" --batch-all-objects' ran
2.06 ± 0.02 times faster than 'git.compile -c pack.readReverseIndex=true cat-file --batch-check="%(objectsize:disk)" --batch-all-objects'
Luckily, the results when running `git cat-file` with `--unordered` are
closer together:
$ hyperfine -L v false,true 'git.compile -c pack.readReverseIndex={v} cat-file --unordered --batch-check="%(objectsize:disk)" --batch-all-objects'
Benchmark 1: git.compile -c pack.readReverseIndex=false cat-file --unordered --batch-check="%(objectsize:disk)" --batch-all-objects
Time (mean ± σ): 5.066 s ± 0.105 s [User: 4.792 s, System: 0.274 s]
Range (min … max): 4.943 s … 5.220 s 10 runs
Benchmark 2: git.compile -c pack.readReverseIndex=true cat-file --unordered --batch-check="%(objectsize:disk)" --batch-all-objects
Time (mean ± σ): 6.193 s ± 0.069 s [User: 5.937 s, System: 0.255 s]
Range (min … max): 6.145 s … 6.356 s 10 runs
Summary
'git.compile -c pack.readReverseIndex=false cat-file --unordered --batch-check="%(objectsize:disk)" --batch-all-objects' ran
1.22 ± 0.03 times faster than 'git.compile -c pack.readReverseIndex=true cat-file --unordered --batch-check="%(objectsize:disk)" --batch-all-objects'
Because the equilibrium point between these two is highly machine- and
repository-dependent, allow users to configure whether or not they will
read any ".rev" file(s) with this configuration knob.
[^1]: Generating a reverse index in memory takes O(N) time (where N is
the number of objects in the repository), since we use a radix sort.
Reading an entry from an on-disk ".rev" file is slower since each
operation is bound by disk I/O instead of memory I/O.
In order to compute the on-disk size of a packed object, we need to
find the offset of our object, and the adjacent object (the on-disk
size difference of these two). Finding the first offset requires a
binary search. Finding the latter involves a single .rev lookup.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Acked-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When no merge.tool or diff.tool is configured or manually selected, the
selection of a default tool is sensitive to the DISPLAY variable; in a
GUI session a gui-specific tool will be proposed if found, and
otherwise a terminal-based one. This "GUI-optimizing" behavior is
important because a GUI can make a huge difference to a user's ability
to understand and correctly complete a non-trivial conflicting merge.
Some time ago the merge.guitool and diff.guitool config options were
introduced to enable users to configure both a GUI tool, and a non-GUI
tool (with fallback if no GUI tool configured), in the same environment.
Unfortunately, the --gui argument introduced to support the selection of
the guitool is still explicit. When using configured tools, there is no
equivalent of the no-tool-configured "propose a GUI tool if we are in a GUI
environment" behavior.
As proposed in <xmqqmtb8jsej.fsf@gitster.g>, introduce new configuration
options, difftool.guiDefault and mergetool.guiDefault, supporting a special
value "auto" which causes the corresponding tool or guitool to be selected
depending on the presence of a non-empty DISPLAY value. Also support "true"
to say "default to the guitool (unless --no-gui is passed on the
commandline)", and "false" as the previous default behavior when these new
configuration options are not specified.
Signed-off-by: Tao Klerks <tao@klerks.biz>
Acked-by: David Aguilar <davvid@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Streamline --rebase-merges command line option handling and
introduce rebase.merges configuration variable.
* ah/rebase-merges-config:
rebase: add a config option for --rebase-merges
rebase: deprecate --rebase-merges=""
rebase: add documentation and test for --no-rebase-merges
The purpose of the new option is to accommodate users who would like
--rebase-merges to be on by default and to facilitate turning on
--rebase-merges by default without configuration in a future version of
Git.
Name the new option rebase.rebaseMerges, even though it is a little
redundant, for consistency with the name of the command line option and
to be clear when scrolling through values in the [rebase] section of
.gitconfig.
Support setting rebase.rebaseMerges to the nonspecific value "true" for
users who don't need to or don't want to learn about the difference
between rebase-cousins and no-rebase-cousins.
Make --rebase-merges without an argument on the command line override
any value of rebase.rebaseMerges in the configuration, for consistency
with other command line flags with optional arguments that have an
associated config option.
Signed-off-by: Alex Henrie <alexhenrie24@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git format-patch" honors the src/dst prefixes set to nonstandard
values with configuration variables like "diff.noprefix", causing
receiving end of the patch that expects the standard -p1 format to
break. Teach "format-patch" to ignore end-user configuration and
always use the standard prefixes.
This is a backward compatibility breaking change.
* jk/format-patch-ignore-noprefix:
rebase: prefer --default-prefix to --{src,dst}-prefix for format-patch
format-patch: add format.noprefix option
format-patch: do not respect diff.noprefix
diff: add --default-prefix option
t4013: add tests for diff prefix options
diff: factor out src/dst prefix setup
The previous commit dropped support for diff.noprefix in format-patch.
While this will do the right thing in most cases (where sending patches
without a prefix was an accidental side effect of the sender preferring
to see their local patches without prefixes), it left no good option for
a project or workflow where you really do want to send patches without
prefixes. You'd be stuck using "--no-prefix" for every invocation.
So let's add a config option specific to format-patch that enables this
behavior. That gives people who have such a workflow a way to get what
they want, but makes it hard to accidentally trigger it.
A more backwards-compatible way of doing the transition would be to have
format.noprefix default to diff.noprefix when it's not set. But that
doesn't really help the "accidental" problem; people would have to
manually set format.noprefix=false. And it's unlikely that anybody
really wants format.noprefix=true in the first place. I'm adding it here
mostly as an escape hatch, not because anybody has expressed any
interest in it.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The user might not necessarily know why ff only was configured, maybe an
admin did it, or the installer (Git for Windows), or perhaps they just
followed some online advice.
This can happen not only on pull.ff=only, but merge.ff=only too.
Even worse if the user has configured pull.rebase=false and
merge.ff=only, because in those cases a diverging merge will constantly
keep failing. There's no trivial way to get out of this other than
`git merge --no-ff`.
Let's not assume our users are experts in git who completely understand
all their configurations.
Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
Acked-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This document only explains PGP signatures, but Git now supports X.509
signatures as of 1e7adb9756 (gpg-interface: introduce new signature
format "x509" using gpgsm, 2018-07-17), and SSH signatures as of
29b315778e (ssh signing: add ssh key format and signing code,
2021-09-10).
Additionally, explain that these signature formats are controlled
`gpg.format`, linking to its documentation, and explain in said
`gpg.format` documentation that the underlying signature format is
documented in signature-format.txt.
Signed-off-by: Gwyneth Morgan <gwymor@tilde.club>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When a lower precedence configuration file (e.g. /etc/gitconfig)
defines format.attach in any way, there was no way to disable it in
a more specific configuration file (e.g. $HOME/.gitconfig).
Change the behaviour of setting it to an empty string. It used to
mean that the result is still a multipart message with only dashes
used as a multi-part separator, but now it resets the setting to
the default (which would be to give an inline patch, unless other
command line options are in effect).
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Finally retire the scripted "git add -p/-i" implementation and have
everybody use the one reimplemented in C.
* ab/retire-scripted-add-p:
docs & comments: replace mentions of "git-add--interactive.perl"
add API: remove run_add_interactive() wrapper function
add: remove "add.interactive.useBuiltin" & Perl "git add--interactive"
The bundle-URI subsystem adds support for creation-token heuristics
to help incremental fetches.
* ds/bundle-uri-5:
bundle-uri: test missing bundles with heuristic
bundle-uri: store fetch.bundleCreationToken
fetch: fetch from an external bundle URI
bundle-uri: drop bundle.flag from design doc
clone: set fetch.bundleURI if appropriate
bundle-uri: download in creationToken order
bundle-uri: parse bundle.<id>.creationToken values
bundle-uri: parse bundle.heuristic=creationToken
t5558: add tests for creationToken heuristic
bundle: verify using check_connected()
bundle: test unbundling with incomplete history
Since [1] first released with Git v2.37.0 the built-in version of "add
-i" has been the default. That built-in implementation was added in
[2], first released with Git v2.25.0.
At this point enough time has passed to allow for finding any
remaining bugs in this new implementation, so let's remove the
fallback code.
As with similar migrations for "stash"[3] and "rebase"[4] we're
keeping a mention of "add.interactive.useBuiltin" in the
documentation, but adding a warning() to notify any outstanding users
that the built-in is now the default. As with [5] and [6] we should
follow-up in the future and eventually remove that warning.
1. 0527ccb1b5 (add -i: default to the built-in implementation,
2021-11-30)
2. f83dff60a7 (Start to implement a built-in version of `git add
--interactive`, 2019-11-13)
3. 8a2cd3f512 (stash: remove the stash.useBuiltin setting,
2020-03-03)
4. d03ebd411c (rebase: remove the rebase.useBuiltin setting,
2019-03-18)
5. deeaf5ee07 (stash: remove documentation for `stash.useBuiltin`,
2022-01-27)
6. 9bcde4d531 (rebase: remove transitory rebase.useBuiltin setting &
env, 2021-03-23)
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When a bundle list specifies the "creationToken" heuristic, the Git
client downloads the list and then starts downloading bundles in
descending creationToken order. This process stops as soon as all
downloaded bundles can be applied to the repository (because all
required commits are present in the repository or in the downloaded
bundles).
When checking the same bundle list twice, this strategy requires
downloading the bundle with the maximum creationToken again, which is
wasteful. The creationToken heuristic promises that the client will not
have a use for that bundle if its creationToken value is at most the
previous creationToken value.
To prevent these wasteful downloads, create a fetch.bundleCreationToken
config setting that the Git client sets after downloading bundles. This
value allows skipping that maximum bundle download when this config
value is the same value (or larger).
To test that this works correctly, we can insert some "duplicate"
fetches into existing tests and demonstrate that only the bundle list is
downloaded.
The previous logic for downloading bundles by creationToken worked even
if the bundle list was empty, but now we have logic that depends on the
first entry of the list. Terminate early in the (non-sensical) case of
an empty bundle list.
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Bundle providers may organize their bundle lists in a way that is
intended to improve incremental fetches, not just initial clones.
However, they do need to state that they have organized with that in
mind, or else the client will not expect to save time by downloading
bundles after the initial clone. This is done by specifying a
bundle.heuristic value.
There are two types of bundle lists: those at a static URI and those
that are advertised from a Git remote over protocol v2.
The new fetch.bundleURI config value applies for static bundle URIs that
are not advertised over protocol v2. If the user specifies a static URI
via 'git clone --bundle-uri', then Git can set this config as a reminder
for future 'git fetch' operations to check the bundle list before
connecting to the remote(s).
For lists provided over protocol v2, we will want to take a different
approach and create a property of the remote itself by creating a
remote.<id>.* type config key. That is not implemented in this change.
Later changes will update 'git fetch' to consume this option.
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The bundle.heuristic value communicates that the bundle list is
organized to make use of the bundle.<id>.creationToken values that may
be provided in the bundle list. Those values will create a total order
on the bundles, allowing the Git client to download them in a specific
order and even remember previously-downloaded bundles by storing the
maximum creation token value.
Before implementing any logic that parses or uses the
bundle.<id>.creationToken values, teach Git to parse the
bundle.heuristic value from a bundle list. We can use 'test-tool
bundle-uri' to print the heuristic value and verify that the parsing
works correctly.
As an extra precaution, create the internal 'heuristics' array to be a
list of (enum, string) pairs so we can iterate through the array entries
carefully, regardless of the enum values.
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Introduce an optional configuration to allow the trailing hash that
protects the index file from bit flipping.
* ds/omit-trailing-hash-in-index:
features: feature.manyFiles implies fast index writes
test-lib-functions: add helper for trailing hash
read-cache: add index.skipHash config option
hashfile: allow skipping the hash function
The recent addition of the index.skipHash config option allows index
writes to speed up by skipping the hash computation for the trailing
checksum. This is particularly critical for repositories with many files
at HEAD, so add this config option to two cases where users in that
scenario may opt-in to such behavior:
1. The feature.manyFiles config option enables some options that are
helpful for repositories with many files at HEAD.
2. 'scalar register' and 'scalar reconfigure' set config options that
optimize for large repositories.
In both of these cases, set index.skipHash=true to gain this
speedup. Add tests that demonstrate the proper way that
index.skipHash=true can override feature.manyFiles=true.
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The previous change allowed skipping the hashing portion of the
hashwrite API, using it instead as a buffered write API. Disabling the
hashwrite can be particularly helpful when the write operation is in a
critical path.
One such critical path is the writing of the index. This operation is so
critical that the sparse index was created specifically to reduce the
size of the index to make these writes (and reads) faster.
This trade-off between file stability at rest and write-time performance
is not easy to balance. The index is an interesting case for a couple
reasons:
1. Writes block users. Writing the index takes place in many user-
blocking foreground operations. The speed improvement directly
impacts their use. Other file formats are typically written in the
background (commit-graph, multi-pack-index) or are super-critical to
correctness (pack-files).
2. Index files are short lived. It is rare that a user leaves an index
for a long time with many staged changes. Outside of staged changes,
the index can be completely destroyed and rewritten with minimal
impact to the user.
Following a similar approach to one used in the microsoft/git fork [1],
add a new config option (index.skipHash) that allows disabling this
hashing during the index write. The cost is that we can no longer
validate the contents for corruption-at-rest using the trailing hash.
[1] 21fed2d914
We load this config from the repository config given by istate->repo,
with a fallback to the_repository if it is not set.
While older Git versions will not recognize the null hash as a special
case, the file format itself is still being met in terms of its
structure. Using this null hash will still allow Git operations to
function across older versions.
The one exception is 'git fsck' which checks the hash of the index file.
This used to be a check on every index read, but was split out to just
the index in a33fc72fe9 (read-cache: force_verify_index_checksum,
2017-04-14) and released first in Git 2.13.0. Document the versions that
relaxed these restrictions, with the optimistic expectation that this
change will be included in Git 2.40.0.
Here, we disable this check if the trailing hash is all zeroes. We add a
warning to the config option that this may cause undesirable behavior
with older Git versions.
As a quick comparison, I tested 'git update-index --force-write' with
and without index.skipHash=true on a copy of the Linux kernel
repository.
Benchmark 1: with hash
Time (mean ± σ): 46.3 ms ± 13.8 ms [User: 34.3 ms, System: 11.9 ms]
Range (min … max): 34.3 ms … 79.1 ms 82 runs
Benchmark 2: without hash
Time (mean ± σ): 26.0 ms ± 7.9 ms [User: 11.8 ms, System: 14.2 ms]
Range (min … max): 16.3 ms … 42.0 ms 69 runs
Summary
'without hash' ran
1.78 ± 0.76 times faster than 'with hash'
These performance benefits are substantial enough to allow users the
ability to opt-in to this feature, even with the potential confusion
with older 'git fsck' versions.
Test this new config option, both at a command-line level and within a
submodule. The confirmation is currently limited to confirm that 'git
fsck' does not complain about the index. Future updates will make this
test more robust.
It is critical that this test is placed before the test_index_version
tests, since those tests obliterate the .git/config file and hence lose
the setting from GIT_TEST_DEFAULT_HASH, if set.
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git format-patch" learned to honor format.mboxrd even when sending
patches to the standard output stream,
* ew/format-patch-mboxrd:
format-patch: support format.mboxrd with --stdout
Bundle URIs part 4.
* ds/bundle-uri-4:
clone: unbundle the advertised bundles
bundle-uri: download bundles from an advertised list
bundle-uri: allow relative URLs in bundle lists
strbuf: introduce strbuf_strip_file_from_path()
bundle-uri: serve bundle.* keys from config
bundle-uri client: add helper for testing server
transport: rename got_remote_heads
bundle-uri client: add boolean transfer.bundleURI setting
clone: request the 'bundle-uri' command when available
t: create test harness for 'bundle-uri' command
protocol v2: add server-side "bundle-uri" skeleton
mboxrd is a more robust output format when used with --stdout
and needs more exposure. Introducing this config knob lets
users choose the more robust format for all their --stdout
uses.
Relying on --pretty=mboxrd and including all of pretty-formats.txt
in the `git format-patch' documentation would likely be
confusing to users. Furthermore, this setting is useful across
multiple invocations. So introduce `format.mboxrd' as a boolean
configuration knob that changes the default --pretty=email format
to --pretty=mboxrd when (and only when) --stdout is in use.
Signed-off-by: Eric Wong <e@80x24.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The yet-to-be introduced client support for bundle-uri will always
fall back on a full clone, but we'd still like to be able to ignore a
server's bundle-uri advertisement entirely.
The new transfer.bundleURI config option defaults to 'false', but a user
can set it to 'true' to enable checking for bundle URIs from the origin
Git server using protocol v2.
Co-authored-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Git learned pushing submodules without pushing the superproject by
the user specifying --recurse-submodules=only through 6c656c3fe4
("submodules: add RECURSE_SUBMODULES_ONLY value", 2016-12-20) and
225e8bf778 ("push: add option to push only submodules", 2016-12-20).
For users who use this feature regularly, it is desirable to have an
equivalent configuration.
It turns out that such a configuration (push.recurseSubmodules=only) is
already supported, even though it is neither documented nor mentioned
in the commit messages, due to the way the --recurse-submodules=only
feature was implemented (a function used to parse --recurse-submodules
was updated to support "only", but that same function is used to parse
push.recurseSubmodules too). What is left is to document it and test it,
which is what this commit does.
There is a possible point of confusion when recursing into a submodule
that itself has the push.recurseSubmodules=only configuration, because
if a repository has only its submodules pushed and not itself, its
superproject can never be pushed. Therefore, treat such configurations
as being "on-demand", and print a warning message.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
The glossary entries for "commit-graph file" and "reachability
bitmap" have been added.
* po/glossary-around-traversal:
glossary: add reachability bitmap description
glossary: add "commit graph" description
doc: use 'object database' not ODB or abbreviation
doc: use "commit-graph" hyphenation consistently
Enable gc.cruftpacks by default for those who opt into
feature.experimental setting.
* es/mark-gc-cruft-as-experimental:
config: let feature.experimental imply gc.cruftPacks=true
gc: add tests for --cruft and friends
Define the logical elements of a "bundle list", data structure to
store them in-core, format to transfer them, and code to parse
them.
* ds/bundle-uri-3:
bundle-uri: suppress stderr from remote-https
bundle-uri: quiet failed unbundlings
bundle: add flags to verify_bundle()
bundle-uri: fetch a list of bundles
bundle: properly clear all revision flags
bundle-uri: limit recursion depth for bundle lists
bundle-uri: parse bundle list in config format
bundle-uri: unit test "key=value" parsing
bundle-uri: create "key=value" line parsing
bundle-uri: create base key-value pair parsing
bundle-uri: create bundle_list struct and helpers
bundle-uri: use plain string in find_temp_filename()
Note, historical release notes have not been updated.
Signed-off-by: Philip Oakley <philipoakley@iee.email>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
We are interested in exploring whether gc.cruftPacks=true should become
the default value.
To determine whether it is safe to do so, let's encourage more users to
try it out.
Users who have set feature.experimental=true have already volunteered to
try new and possibly-breaking config changes, so let's try this new
default with that set of users.
Signed-off-by: Emily Shaffer <emilyshaffer@google.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The documentation lacks mention of specific <msg-id> that are supported.
While git-help --config will display a list of these options, often
developers' first instinct is to consult the git docs to find valid
config values.
Add a list of fsck error messages, and link to it from the git-fsck
documentation.
Signed-off-by: John Cai <johncai86@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
By default, use of fsmonitor on a repository on networked
filesystem is disabled. Add knobs to make it workable on macOS.
* ed/fsmonitor-on-networked-macos:
fsmonitor: fix leak of warning message
fsmonitor: add documentation for allowRemote and socketDir options
fsmonitor: check for compatability before communicating with fsmonitor
fsmonitor: deal with synthetic firmlinks on macOS
fsmonitor: avoid socket location check if using hook
fsmonitor: relocate socket file if .git directory is remote
fsmonitor: refactor filesystem checks to common interface
