| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Doc fix.
* dp/submodule-doc-markup-fix:
config.txt: fix formatting of submodule.alternateErrorStrategy section
|
|
Add missing `::` after the title.
Signed-off-by: David Pursehouse <dpursehouse@collab.net>
Acked-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The "core.logAllRefUpdates" that used to be boolean has been
enhanced to take 'always' as well, to record ref updates to refs
other than the ones that are expected to be updated (i.e. branches,
remote-tracking branches and notes).
* cw/log-updates-for-all-refs-really:
doc: add note about ignoring '--no-create-reflog'
update-ref: add test cases for bare repository
refs: add option core.logAllRefUpdates = always
config: add markup to core.logAllRefUpdates doc
|
|
Some people feel the default set of colors used by "git log --graph"
rather limiting. A mechanism to customize the set of colors has
been introduced.
* nd/log-graph-configurable-colors:
document behavior of empty color name
color_parse_mem: allow empty color spec
log --graph: customize the graph lines with config log.graphColors
color.c: trim leading spaces in color_parse_mem()
color.c: fix color_parse_mem() with value_len == 0
|
|
Commit 55cccf4bb (color_parse_mem: allow empty color spec,
2017-02-01) clearly defined the behavior of an empty color
config variable. Let's document that, and give a hint about
why it might be useful.
It's important not to say that it makes the item uncolored,
because it doesn't. It just sets no attributes, which means
that any previous attributes continue to take effect.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
When core.logallrefupdates is true, we only create a new reflog for refs
that are under certain well-known hierarchies. The reason is that we
know that some hierarchies (like refs/tags) are not meant to change, and
that unknown hierarchies might not want reflogs at all (e.g., a
hypothetical refs/foo might be meant to change often and drop old
history immediately).
However, sometimes it is useful to override this decision and simply log
for all refs, because the safety and audit trail is more important than
the performance implications of keeping the log around.
This patch introduces a new "always" mode for the core.logallrefupdates
option which will log updates to everything under refs/, regardless
where in the hierarchy it is (we still will not log things like
ORIG_HEAD and FETCH_HEAD, which are known to be transient).
Based-on-patch-by: Jeff King <peff@peff.net>
Signed-off-by: Cornelius Weig <cornelius.weig@tngtech.com>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Signed-off-by: Cornelius Weig <cornelius.weig@tngtech.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
If you have a 256 colors terminal (or one with true color support), then
the predefined 12 colors seem limited. On the other hand, you don't want
to draw graph lines with every single color in this mode because the two
colors could look extremely similar. This option allows you to hand pick
the colors you want.
Even with standard terminal, if your background color is neither black
or white, then the graph line may match your background and become
hidden. You can exclude your background color (or simply the colors you
hate) with this.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The prereleaseSuffix feature of version comparison that is used in
"git tag -l" did not correctly when two or more prereleases for the
same release were present (e.g. when 2.0, 2.0-beta1, and 2.0-beta2
are there and the code needs to compare 2.0-beta1 and 2.0-beta2).
* sg/fix-versioncmp-with-common-suffix:
versioncmp: generalize version sort suffix reordering
versioncmp: factor out helper for suffix matching
versioncmp: use earliest-longest contained suffix to determine sorting order
versioncmp: cope with common part overlapping with prerelease suffix
versioncmp: pass full tagnames to swap_prereleases()
t7004-tag: add version sort tests to show prerelease reordering issues
t7004-tag: use test_config helper
t7004-tag: delete unnecessary tags with test_when_finished
|
|
The 'versionsort.prereleaseSuffix' configuration variable, as its name
suggests, is supposed to only deal with tagnames with prerelease
suffixes, and allows sorting those prerelease tags in a user-defined
order before the suffixless main release tag, instead of sorting them
simply lexicographically.
However, the previous changes in this series resulted in an
interesting and useful property of version sort:
- The empty string as a configured suffix matches all tagnames,
including tagnames without any suffix, but
- tagnames containing a "real" configured suffix are still ordered
according to that real suffix, because any longer suffix takes
precedence over the empty string.
Exploiting this property we can easily generalize suffix reordering
and specify the order of tags with given suffixes not only before but
even after a main release tag by using the empty suffix to denote the
position of the main release tag, without any algorithm changes:
$ git -c versionsort.prereleaseSuffix=-alpha \
-c versionsort.prereleaseSuffix=-beta \
-c versionsort.prereleaseSuffix="" \
-c versionsort.prereleaseSuffix=-gamma \
-c versionsort.prereleaseSuffix=-delta \
tag -l --sort=version:refname 'v3.0*'
v3.0-alpha1
v3.0-beta1
v3.0
v3.0-gamma1
v3.0-delta1
Since 'versionsort.prereleaseSuffix' is not a fitting name for a
configuration variable to control this more general suffix reordering,
introduce the new variable 'versionsort.suffix'. Still keep the old
configuration variable name as a deprecated alias, though, to avoid
suddenly breaking setups already using it. Ignore the old variable if
both old and new configuration variables are set, but emit a warning
so users will be aware of it and can fix their configuration. Extend
the documentation to describe and add a test to check this more
general behavior.
Note: since the empty suffix matches all tagnames, tagnames with
suffixes not included in the configuration are listed together with
the suffixless main release tag, ordered lexicographically right after
that, i.e. before tags with suffixes listed in the configuration
following the empty suffix.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Recent update to the default abbreviation length that auto-scales
lacked documentation update, which has been corrected.
* jc/abbrev-autoscale-config:
config.abbrev: document the new default that auto-scales
|
|
When the http server gives an incomplete response to a smart-http
rpc call, it could lead to client waiting for a full response that
will never come. Teach the client side to notice this condition
and abort the transfer.
An improvement counterproposal has failed.
cf. <20161114194049.mktpsvgdhex2f4zv@sigill.intra.peff.net>
* dt/smart-http-detect-server-going-away:
upload-pack: optionally allow fetching any sha1
remote-curl: don't hang when a server dies before any output
|
|
Doc update.
* mm/gc-safety-doc:
git-gc.txt: expand discussion of races with other processes
|
|
Doc update on fetching and pushing.
* mm/push-social-engineering-attack-doc:
doc: mention transfer data leaks in more places
|
|
Finer-grained control of what protocols are allowed for transports
during clone/fetch/push have been enabled via a new configuration
mechanism.
* bw/transport-protocol-policy:
http: respect protocol.*.allow=user for http-alternates
transport: add from_user parameter to is_transport_allowed
http: create function to get curl allowed protocols
transport: add protocol policy config option
http: always warn if libcurl version is too old
lib-proto-disable: variable name fix
|
|
We somehow forgot to update the "default is 7" in the
documentation. Also give a way to explicitly ask the auto-scaling
by setting config.abbrev to "auto".
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Transport with dumb http can be fooled into following foreign URLs
that the end user does not intend to, especially with the server
side redirects and http-alternates mechanism, which can lead to
security issues. Tighten the redirection and make it more obvious
to the end user when it happens.
* jk/http-walker-limit-redirect-2.9:
http: treat http-alternates like redirects
http: make redirects more obvious
remote-curl: rename shadowed options variable
http: always update the base URL for redirects
http: simplify update_url_from_redirect
|
|
Previously the `GIT_ALLOW_PROTOCOL` environment variable was used to
specify a whitelist of protocols to be used in clone/fetch/push
commands. This patch introduces new configuration options for more
fine-grained control for allowing/disallowing protocols. This also has
the added benefit of allowing easier construction of a protocol
whitelist on systems where setting an environment variable is
non-trivial.
Now users can specify a policy to be used for each type of protocol via
the 'protocol.<name>.allow' config option. A default policy for all
unconfigured protocols can be set with the 'protocol.allow' config
option. If no user configured default is made git will allow known-safe
protocols (http, https, git, ssh, file), disallow known-dangerous
protocols (ext), and have a default policy of `user` for all other
protocols.
The supported policies are `always`, `never`, and `user`. The `user`
policy can be used to configure a protocol to be usable when explicitly
used by a user, while disallowing it for commands which run
clone/fetch/push commands without direct user intervention (e.g.
recursive initialization of submodules). Commands which can potentially
clone/fetch/push from untrusted repositories without user intervention
can export `GIT_PROTOCOL_FROM_USER` with a value of '0' to prevent
protocols configured to the `user` policy from being used.
Fix remote-ext tests to use the new config to allow the ext
protocol to be tested.
Based on a patch by Jeff King <peff@peff.net>
Signed-off-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
When comparing tagnames, it is possible that a tagname contains more
than one of the configured prerelease suffixes around the first
different character. After fixing a bug in the previous commit such a
tagname is sorted according to the contained suffix which comes first
in the configuration. This is, however, not quite the right thing to
do in the following corner cases:
1. $ git -c versionsort.suffix=-bar
-c versionsort.suffix=-foo-baz
-c versionsort.suffix=-foo-bar
tag -l --sort=version:refname 'v1*'
v1.0-foo-bar
v1.0-foo-baz
The suffix of the tagname 'v1.0-foo-bar' is clearly '-foo-bar',
so it should be listed last. However, as it also contains '-bar'
around the first different character, it is listed first instead,
because that '-bar' suffix comes first the configuration.
2. One of the configured suffixes starts with the other:
$ git -c versionsort.prereleasesuffix=-pre \
-c versionsort.prereleasesuffix=-prerelease \
tag -l --sort=version:refname 'v2*'
v2.0-prerelease1
v2.0-pre1
v2.0-pre2
Here the tagname 'v2.0-prerelease1' should be the last. When
comparing 'v2.0-pre1' and 'v2.0-prerelease1' the first different
characters are '1' and 'r', respectively. Since this first
different character must be part of the configured suffix, the
'-pre' suffix is not recognized in the first tagname. OTOH, the
'-prerelease' suffix is properly recognized in
'v2.0-prerelease1', thus it is listed first.
Improve version sort in these corner cases, and
- look for a configured prerelease suffix containing the first
different character or ending right before it, so the '-pre'
suffixes are recognized in case (2). This also means that
when comparing tagnames 'v2.0-pre1' and 'v2.0-pre2',
swap_prereleases() would find the '-pre' suffix in both, but then
it will return "undecided" and the caller will do the right thing
by sorting based in '1' and '2'.
- If the tagname contains more than one suffix, then give precedence
to the contained suffix that starts at the earliest offset in the
tagname to address (1).
- If there are more than one suffixes starting at that earliest
position, then give precedence to the longest of those suffixes,
thus ensuring that in (2) the tagname 'v2.0-prerelease1' won't be
sorted based on the '-pre' suffix.
Add tests for these corner cases and adjust the documentation
accordingly.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Version sort with prerelease reordering sometimes puts tagnames in the
wrong order, when the common part of two compared tagnames overlaps
with the leading character(s) of one or more configured prerelease
suffixes. Note the position of "v2.1.0-beta-1":
$ git -c versionsort.prereleaseSuffix=-beta \
tag -l --sort=version:refname v2.1.*
v2.1.0-beta-2
v2.1.0-beta-3
v2.1.0
v2.1.0-RC1
v2.1.0-RC2
v2.1.0-beta-1
v2.1.1
v2.1.2
The reason is that when comparing a pair of tagnames, first
versioncmp() looks for the first different character in a pair of
tagnames, and then the swap_prereleases() helper function looks for a
configured prerelease suffix _starting at_ that character. Thus, when
in the above example the sorting algorithm happens to compare the
tagnames "v2.1.0-beta-1" and "v2.1.0-RC2", swap_prereleases() tries to
match the suffix "-beta" against "beta-1" to no avail, and the two
tagnames erroneously end up being ordered lexicographically.
To fix this issue change swap_prereleases() to look for configured
prerelease suffixes _containing_ the position of that first different
character.
Care must be taken, when a configured suffix is longer than the
tagnames' common part up to the first different character, to avoid
reading memory before the beginning of the tagnames. Add a test that
uses an exceptionally long prerelease suffix to check for this, in the
hope that in case of a regression the illegal memory access causes a
segfault in 'git tag' on one of the commonly used platforms (the test
happens to pass successfully on my Linux system with the safety check
removed), or at least makes valgrind complain.
Under some circumstances it's possible that more than one prerelease
suffixes can be found in the same tagname around that first different
character. With this simple bugfix patch such a tagname is sorted
according to the contained suffix that comes first in the
configuration for now. This is less than ideal in some cases, and the
following patch will take care of those.
Reported-by: Leho Kraav <leho@conversionready.com>
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
We instruct curl to always follow HTTP redirects. This is
convenient, but it creates opportunities for malicious
servers to create confusing situations. For instance,
imagine Alice is a git user with access to a private
repository on Bob's server. Mallory runs her own server and
wants to access objects from Bob's repository.
Mallory may try a few tricks that involve asking Alice to
clone from her, build on top, and then push the result:
1. Mallory may simply redirect all fetch requests to Bob's
server. Git will transparently follow those redirects
and fetch Bob's history, which Alice may believe she
got from Mallory. The subsequent push seems like it is
just feeding Mallory back her own objects, but is
actually leaking Bob's objects. There is nothing in
git's output to indicate that Bob's repository was
involved at all.
The downside (for Mallory) of this attack is that Alice
will have received Bob's entire repository, and is
likely to notice that when building on top of it.
2. If Mallory happens to know the sha1 of some object X in
Bob's repository, she can instead build her own history
that references that object. She then runs a dumb http
server, and Alice's client will fetch each object
individually. When it asks for X, Mallory redirects her
to Bob's server. The end result is that Alice obtains
objects from Bob, but they may be buried deep in
history. Alice is less likely to notice.
Both of these attacks are fairly hard to pull off. There's a
social component in getting Mallory to convince Alice to
work with her. Alice may be prompted for credentials in
accessing Bob's repository (but not always, if she is using
a credential helper that caches). Attack (1) requires a
certain amount of obliviousness on Alice's part while making
a new commit. Attack (2) requires that Mallory knows a sha1
in Bob's repository, that Bob's server supports dumb http,
and that the object in question is loose on Bob's server.
But we can probably make things a bit more obvious without
any loss of functionality. This patch does two things to
that end.
First, when we encounter a whole-repo redirect during the
initial ref discovery, we now inform the user on stderr,
making attack (1) much more obvious.
Second, the decision to follow redirects is now
configurable. The truly paranoid can set the new
http.followRedirects to false to avoid any redirection
entirely. But for a more practical default, we will disallow
redirects only after the initial ref discovery. This is
enough to thwart attacks similar to (2), while still
allowing the common use of redirects at the repository
level. Since c93c92f30 (http: update base URLs when we see
redirects, 2013-09-28) we re-root all further requests from
the redirect destination, which should generally mean that
no further redirection is necessary.
As an escape hatch, in case there really is a server that
needs to redirect individual requests, the user can set
http.followRedirects to "true" (and this can be done on a
per-server basis via http.*.followRedirects config).
Reported-by: Jann Horn <jannh@google.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
It seems a little silly to do a reachabilty check in the case where we
trust the user to access absolutely everything in the repository.
Also, it's racy in a distributed system -- perhaps one server
advertises a ref, but another has since had a force-push to that ref,
and perhaps the two HTTP requests end up directed to these different
servers.
Signed-off-by: David Turner <dturner@twosigma.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
In general, "git gc" may delete objects that another concurrent process
is using but hasn't created a reference to. Git has some mitigations,
but they fall short of a complete solution. Document this in the
git-gc(1) man page and add a reference from the documentation of the
gc.pruneExpire config variable.
Based on a write-up by Jeff King:
http://marc.info/?l=git&m=147922960131779&w=2
Signed-off-by: Matt McCutchen <matt@mattmccutchen.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The "SECURITY" section of the gitnamespaces(7) man page described two
ways for a client to steal data from a server that wasn't intended to be
shared. Similar attacks can be performed by a server on a client, so
adapt the section to cover both directions and add it to the
git-fetch(1), git-pull(1), and git-push(1) man pages. Also add
references to this section from the documentation of server
configuration options that attempt to control data leakage but may not
be fully effective.
Signed-off-by: Matt McCutchen <matt@mattmccutchen.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Documentation fix.
* jk/rebase-config-insn-fmt-docfix:
doc: fix missing "::" in config list
|
|
The rebase.instructionFormat option is missing its "::" to
tell AsciiDoc that it's a list entry. As a result, the
option name gets lumped into the description in one big
paragraph.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The "submodule.<name>.path" stored in .gitmodules is never copied
to .git/config and such a key in .git/config has no meaning, but
the documentation described it and submodule.<name>.url next to
each other as if both belong to .git/config. This has been fixed.
* sb/submodule-config-doc-drop-path:
documentation: improve submodule.<name>.{url, path} description
|
|
Unlike the url variable a user cannot override the the path variable,
as it is part of the content together with the gitlink at the given
path. To avoid confusion do not mention the .path variable in the config
section and rely on the documentation provided in gitmodules[5].
Enhance the description of submodule.<name>.url and mention its two use
cases separately.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
In recent versions of cURL, GSSAPI credential delegation is
disabled by default due to CVE-2011-2192; introduce a configuration
to selectively allow enabling this.
* ps/http-gssapi-cred-delegation:
http: control GSSAPI credential delegation
|
|
Delegation of credentials is disabled by default in libcurl since
version 7.21.7 due to security vulnerability CVE-2011-2192. Which
makes troubles with GSS/kerberos authentication when delegation
of credentials is required. This can be changed with option
CURLOPT_GSSAPI_DELEGATION in libcurl with set expected parameter
since libcurl version 7.22.0.
This patch provides new configuration variable http.delegation
which corresponds to curl parameter "--delegation" (see man 1 curl).
The following values are supported:
* none (default).
* policy
* always
Signed-off-by: Petr Stodulka <pstodulk@redhat.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Documentation for individual configuration variables to control use
of color (like `color.grep`) said that their default value is
'false', instead of saying their default is taken from `color.ui`.
When we updated the default value for color.ui from 'false' to
'auto' quite a while ago, all of them broke. This has been
corrected.
* mm/config-color-ui-default-to-auto:
Documentation/config: default for color.* is color.ui
|
|
"git gc --aggressive" used to limit the delta-chain length to 250,
which is way too deep for gaining additional space savings and is
detrimental for runtime performance. The limit has been reduced to
50.
* jk/reduce-gc-aggressive-depth:
gc: default aggressive depth to 50
|
|
Since 4c7f181 (make color.ui default to 'auto', 2013-06-10), the
default for color.* when nothing is set is 'auto' and we still claimed
that the default was 'false'. Be more precise by saying explicitly
that the default is to follow color.ui, and recall that the default is
'auto' to avoid one indirection for the reader.
Signed-off-by: Matthieu Moy <Matthieu.Moy@imag.fr>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
"git clone --resurse-submodules --reference $path $URL" is a way to
reduce network transfer cost by borrowing objects in an existing
$path repository when cloning the superproject from $URL; it
learned to also peek into $path for presense of corresponding
repositories of submodules and borrow objects from there when able.
* sb/submodule-clone-rr:
clone: recursive and reference option triggers submodule alternates
clone: implement optional references
clone: clarify option_reference as required
clone: factor out checking for an alternate path
submodule--helper update-clone: allow multiple references
submodule--helper module-clone: allow multiple references
t7408: merge short tests, factor out testing method
t7408: modernize style
|
|
"git merge" with renormalization did not work well with
merge-recursive, due to "safer crlf" conversion kicking in when it
shouldn't.
* jc/renormalize-merge-kill-safer-crlf:
merge: avoid "safer crlf" during recording of merge results
convert: unify the "auto" handling of CRLF
|
|
Receive-pack feeds its input to either index-pack or
unpack-objects, which will happily accept as many bytes as
a sender is willing to provide. Let's allow an arbitrary
cutoff point where we will stop writing bytes to disk.
Cleaning up what has already been written to disk is a
related problem that is not addressed by this patch.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
When `--recursive` and `--reference` is given, it is reasonable to
expect that the submodules are created with references to the submodules
of the given alternate for the superproject.
An initial attempt to do this was presented to the mailing list, which
used flags that are passed around ("--super-reference") that instructed
the submodule clone to look for a reference in the submodules of the
referenced superproject. This is not well thought out, as any further
`submodule update` should also respect the initial setup.
When a new submodule is added to the superproject and the alternate
of the superproject does not know about that submodule yet, we rather
error out informing the user instead of being unclear if we did or did
not use a submodules alternate.
To solve this problem introduce new options that store the configuration
for what the user wanted originally.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
This commit message is long and has lots of background and
numbers. The summary is: the current default of 250 doesn't
save much space, and costs CPU. It's not a good tradeoff.
Read on for details.
The "--aggressive" flag to git-gc does three things:
1. use "-f" to throw out existing deltas and recompute from
scratch
2. use "--window=250" to look harder for deltas
3. use "--depth=250" to make longer delta chains
Items (1) and (2) are good matches for an "aggressive"
repack. They ask the repack to do more computation work in
the hopes of getting a better pack. You pay the costs during
the repack, and other operations see only the benefit.
Item (3) is not so clear. Allowing longer chains means fewer
restrictions on the deltas, which means potentially finding
better ones and saving some space. But it also means that
operations which access the deltas have to follow longer
chains, which affects their performance. So it's a tradeoff,
and it's not clear that the tradeoff is even a good one.
The existing "250" numbers for "--aggressive" come
originally from this thread:
http://public-inbox.org/git/alpine.LFD.0.9999.0712060803430.13796@woody.linux-foundation.org/
where Linus says:
So when I said "--depth=250 --window=250", I chose those
numbers more as an example of extremely aggressive
packing, and I'm not at all sure that the end result is
necessarily wonderfully usable. It's going to save disk
space (and network bandwidth - the delta's will be re-used
for the network protocol too!), but there are definitely
downsides too, and using long delta chains may
simply not be worth it in practice.
There are some numbers in that thread, but they're mostly
focused on the improved window size, and measure the
improvement from --depth=250 and --window=250 together.
E.g.:
http://public-inbox.org/git/9e4733910712062006l651571f3w7f76ce64c6650dff@mail.gmail.com/
talks about the improved run-time of "git-blame", which
comes from the reduced pack size. But most of that reduction
is coming from --window=250, whereas most of the extra costs
come from --depth=250. There's a link in that thread showing
that increasing the depth beyond 50 doesn't seem to help
much with the size:
https://vcscompare.blogspot.com/2008/06/git-repack-parameters.html
but again, no discussion of the timing impact.
In an earlier thread from Ted Ts'o which discussed setting
the non-aggressive default (from 10 to 50):
http://public-inbox.org/git/20070509134958.GA21489%40thunk.org/
we have more numbers, with the conclusion that going past 50
does not help size much, and hurts the speed of normal
operations.
So from that, we might guess that 50 is actually a sweet
spot, even for aggressive, if we interpret aggressive to
"spend time now to make a better pack". It is not clear that
"--depth=250" is actually a better pack. It may be slightly
_smaller_, but it carries a run-time penalty.
Here are some more recent timings I did to verify that. They
show three things:
- the size of the resulting pack (so disk saved to store,
bandwidth saved on clones/fetches)
- the cost of "rev-list --objects --all", which shows the
effect of the delta chains on trees (commits typically
don't delta, and the command doesn't touch the blobs at
all)
- the cost of "log -Sfoo", which will additionally access
each blob
All cases were repacked with "git repack -adf --depth=$d
--window=250" (so basically, what would happen if we tweaked
the "gc --aggressive" default depth).
The timings are all wall-clock best-of-3. The machine itself
has plenty of RAM compared to the repositories (which is
probably typical of most workstations these days), so we're
really measuring CPU usage, as the whole thing will be in
disk cache after the first run.
The core.deltaBaseCacheLimit is at its default of 96MiB.
It's possible that tweaking it would have some impact on the
tests, as some of them (especially "log -S" on a large repo)
are likely to overflow that. But bumping that carries a
run-time memory cost, so for these tests, I focused on what
we could do just with the on-disk pack tradeoffs.
Each test is done for four depths: 250 (the current value),
50 (the current default that tested well previously), 100
(to show something on the larger side, which previous tests
showed was not a good tradeoff), and 10 (the very old
default, which previous tests showed was worse than 50).
Here are the numbers for linux.git:
depth | size | % | rev-list | % | log -Sfoo | %
-------+-------+-------+----------+--------+-----------+-------
250 | 967MB | n/a | 48.159s | n/a | 378.088 | n/a
100 | 971MB | +0.4% | 41.471s | -13.9% | 342.060 | -9.5%
50 | 979MB | +1.2% | 37.778s | -21.6% | 311.040s | -17.7%
10 | 1.1GB | +6.6% | 32.518s | -32.5% | 279.890s | -25.9%
and for git.git:
depth | size | % | rev-list | % | log -Sfoo | %
-------+-------+-------+----------+--------+-----------+-------
250 | 48MB | n/a | 2.215s | n/a | 20.922s | n/a
100 | 49MB | +0.5% | 2.140s | -3.4% | 17.736s | -15.2%
50 | 49MB | +1.7% | 2.099s | -5.2% | 15.418s | -26.3%
10 | 53MB | +9.3% | 2.001s | -9.7% | 12.677s | -39.4%
You can see that that the CPU savings for regular operations improves as we
decrease the depth. The savings are less for "rev-list" on a smaller repository
than they are for blob-accessing operations, or even rev-list on a larger
repository. This may mean that a larger delta cache would help (though setting
core.deltaBaseCacheLimit by itself doesn't).
But we can also see that the space savings are not that great as the depth goes
higher. Saving 5-10% between 10 and 50 is probably worth the CPU tradeoff.
Saving 1% to go from 50 to 100, or another 0.5% to go from 100 to 250 is
probably not.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
"git format-patch" learned format.from configuration variable to
specify the default settings for its "--from" option.
* jt/format-patch-from-config:
format-patch: format.from gives the default for --from
|
|
"git push" and "git clone" learned to give better progress meters
to the end user who is waiting on the terminal.
* jk/push-progress:
receive-pack: send keepalives during quiet periods
receive-pack: turn on connectivity progress
receive-pack: relay connectivity errors to sideband
receive-pack: turn on index-pack resolving progress
index-pack: add flag for showing delta-resolution progress
clone: use a real progress meter for connectivity check
check_connected: add progress flag
check_connected: relay errors to alternate descriptor
check_everything_connected: use a struct with named options
check_everything_connected: convert to argv_array
rev-list: add optional progress reporting
check_everything_connected: always pass --quiet to rev-list
|
|
"git push" learned to accept and pass extra options to the
receiving end so that hooks can read and react to them.
* sb/push-options:
add a test for push options
push: accept push options
receive-pack: implement advertising and receiving push options
push options: {pre,post}-receive hook learns about push options
|
|
This helps users who would prefer format-patch to default to --from,
and makes it easier to change the default in the future.
Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
More mark-up updates to typeset strings that are expected to
literally typed by the end user in fixed-width font.
* mm/doc-tt:
doc: typeset HEAD and variants as literal
CodingGuidelines: formatting HEAD in documentation
doc: typeset long options with argument as literal
doc: typeset '--' as literal
doc: typeset long command-line options as literal
doc: typeset short command-line options as literal
Documentation/git-mv.txt: fix whitespace indentation
|
|
"git merge" with renormalization did not work well with
merge-recursive, due to "safer crlf" conversion kicking in when it
shouldn't.
* jc/renormalize-merge-kill-safer-crlf:
merge: avoid "safer crlf" during recording of merge results
convert: unify the "auto" handling of CRLF
|
|
After a client has sent us the complete pack, we may spend
some time processing the data and running hooks. If the
client asked us to be quiet, receive-pack won't send any
progress data during the index-pack or connectivity-check
steps. And hooks may or may not produce their own progress
output. In these cases, the network connection is totally
silent from both ends.
Git itself doesn't care about this (it will wait forever),
but other parts of the system (e.g., firewalls,
load-balancers, etc) might hang up the connection. So we'd
like to send some sort of keepalive to let the network and
the client side know that we're still alive and processing.
We can use the same trick we did in 05e9515 (upload-pack:
send keepalive packets during pack computation, 2013-09-08).
Namely, we will send an empty sideband data packet every `N`
seconds that we do not relay any stderr data over the
sideband channel. As with 05e9515, this means that we won't
bother sending keepalives when there's actual progress data,
but will kick in when it has been disabled (or if there is a
lull in the progress data).
The concept is simple, but the details are subtle enough
that they need discussing here.
Before the client sends us the pack, we don't want to do any
keepalives. We'll have sent our ref advertisement, and we're
waiting for them to send us the pack (and tell us that they
support sidebands at all).
While we're receiving the pack from the client (or waiting
for it to start), there's no need for keepalives; it's up to
them to keep the connection active by sending data.
Moreover, it would be wrong for us to do so. When we are the
server in the smart-http protocol, we must treat our
connection as half-duplex. So any keepalives we send while
receiving the pack would potentially be buffered by the
webserver. Not only does this make them useless (since they
would not be delivered in a timely manner), but it could
actually cause a deadlock if we fill up the buffer with
keepalives. (It wouldn't be wrong to send keepalives in this
phase for a full-duplex connection like ssh; it's simply
pointless, as it is the client's responsibility to speak).
As soon as we've gotten all of the pack data, then the
client is waiting for us to speak, and we should start
keepalives immediately. From here until the end of the
connection, we send one any time we are not otherwise
sending data.
But there's a catch. Receive-pack doesn't know the moment
we've gotten all the data. It passes the descriptor to
index-pack, who reads all of the data, and then starts
resolving the deltas. We have to communicate that back.
To make this work, we instruct the sideband muxer to enable
keepalives in three phases:
1. In the beginning, not at all.
2. While reading from index-pack, wait for a signal
indicating end-of-input, and then start them.
3. Afterwards, always.
The signal from index-pack in phase 2 has to come over the
stderr channel which the muxer is reading. We can't use an
extra pipe because the portable run-command interface only
gives us stderr and stdout.
Stdout is already used to pass the .keep filename back to
receive-pack. We could also send a signal there, but then we
would find out about it in the main thread. And the
keepalive needs to be done by the async muxer thread (since
it's the one writing sideband data back to the client). And
we can't reliably signal the async thread from the main
thread, because the async code sometimes uses threads and
sometimes uses forked processes.
Therefore the signal must come over the stderr channel,
where it may be interspersed with other random
human-readable messages from index-pack. This patch makes
the signal a single NUL byte. This is easy to parse, should
not appear in any normal stderr output, and we don't have to
worry about any timing issues (like seeing half the signal
bytes in one read(), and half in a subsequent one).
This is a bit ugly, but it's simple to code and should work
reliably.
Another option would be to stop using an async thread for
muxing entirely, and just poll() both stderr and stdout of
index-pack from the main thread. This would work for
index-pack (because we aren't doing anything useful in the
main thread while it runs anyway). But it would make the
connectivity check and the hook muxers much more
complicated, as they need to simultaneously feed the
sub-programs while reading their stderr.
The index-pack phase is the only one that needs this
signaling, so it could simply behave differently than the
other two. That would mean having two separate
implementations of copy_to_sideband (and the keepalive
code), though. And it still doesn't get rid of the
signaling; it just means we can write a nicer message like
"END_OF_INPUT" or something on stdout, since we don't have
to worry about separating it from the stderr cruft.
One final note: this signaling trick is only done with
index-pack, not with unpack-objects. There's no point in
doing it for the latter, because by definition it only kicks
in for a small number of objects, where keepalives are not
as useful (and this conveniently lets us avoid duplicating
the implementation).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Improve the look of the way "git fetch" reports what happened to
each ref that was fetched.
* nd/fetch-ref-summary:
fetch: reduce duplicate in ref update status lines with placeholder
fetch: align all "remote -> local" output
fetch: change flag code for displaying tag update and deleted ref
fetch: refactor ref update status formatting code
git-fetch.txt: document fetch output
|
|
A new configuration variable core.sshCommand has been added to
specify what value for GIT_SSH_COMMAND to use per repository.
* nd/connect-ssh-command-config:
connect: read $GIT_SSH_COMMAND from config file
|
|
The pre/post receive hook may be interested in more information from the
user. This information can be transmitted when both client and server
support the "push-options" capability, which when used is a phase directly
after update commands ended by a flush pkt.
Similar to the atomic option, the server capability can be disabled via
the `receive.advertisePushOptions` config variable. While documenting
this, fix a nit in the `receive.advertiseAtomic` wording.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
More mark-up updates to typeset strings that are expected to
literally typed by the end user in fixed-width font.
* mm/doc-tt:
doc: typeset HEAD and variants as literal
CodingGuidelines: formatting HEAD in documentation
doc: typeset long options with argument as literal
doc: typeset '--' as literal
doc: typeset long command-line options as literal
doc: typeset short command-line options as literal
Documentation/git-mv.txt: fix whitespace indentation
|
|
The output coloring scheme learned two new attributes, italic and
strike, in addition to existing bold, reverse, etc.
* jk/ansi-color:
color: support strike-through attribute
color: support "italic" attribute
color: allow "no-" for negating attributes
color: refactor parse_attr
add skip_prefix_mem helper
doc: refactor description of color format
color: fix max-size comment
|
Junio C Hamano
David Pursehouse
Jeff King
Cornelius Weig
Nguyễn Thái Ngọc Duy
SZEDER Gábor
Brandon Williams
David Turner
Matt McCutchen
Stefan Beller
Petr Stodulka
Matthieu Moy
Josh Triplett