| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
Change-Id: I5e6ce63b23c72f3fc293940e94a00affed294e77
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Since Bazel 0.16, the build is done with an embedded JRE using Java
version 9. This causes "illegal reflective access operation" warnings
during the documentation build, because the version of jruby (9.1.13)
doesn't support Java 9.
Support for Java 9 was added in jruby 9.1.14 [1]. The subsequent
releases also include bug fixes (see [2], [3], [4]).
Note that we don't upgrade to the latest (9.2.0).
Also upgrade asciidoctorj to the latest release, 1.5.7.
[1] http://jruby.org/2017/11/08/jruby-9-1-14-0.html
[2] http://jruby.org/2017/12/07/jruby-9-1-15-0.html
[3] http://jruby.org/2018/02/21/jruby-9-1-16-0.html
[4] http://jruby.org/2018/04/23/jruby-9-1-17-0.html
Bug: Issue 9766
Change-Id: I2f84612905761c4bdfddca807e92f54247ef6f6e
|
|
|
|
|
|
|
|
|
|
| |
When using Amazon's Elasticsearch service, requests fail with "400 Bad
Request" if they are not prefixed with "/".
Ensure that requests are always prefixed with "/".
Bug: Issue 9761
Change-Id: If2a178e85143ba75d7f4f6b982739a9b3f2dc21c
|
|
|
|
|
|
|
|
|
|
|
|
| |
Factor out more variants of these methods that eventually end up in the
one that actually makes the call to the RestClient instance. This will
allow to make any necessary adjustments to the URI or request parameters
in a single place.
At the same time, reorder the arguments to make the order consistent
across methods.
Change-Id: I3ced32e8df6cbd5a5f05f4d3241f84a7e7bebb10
|
|
|
|
| |
Change-Id: Ifddb6d4b6aa615205e60a6dc13388ecf8de9b46e
|
|
|
|
|
|
|
|
| |
During the documentation build, a warning is emitted because the
reference "database.h2.cachesize" does not exist. The actual name
is "database.h2.cacheSize" with upper-case S.
Change-Id: I6481e77a88a2b340197a1369f1aa25973df83417
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In I44ca2ecfea6 native http_file was replaced with Starlark rule. During
this migration the original file name was lost and hard coded to file
named "downloaded". The closure_js_library expects files with .js suffix
as source files, so that we had to add an intermediate rename step to
make it work gain.
In context of this feature request: [1] downloaded_file_path was added
to the http_file rule: [2] so that we can use it now and can remove the
intermediate renaming step.
The aformentioned fix was included in 0.17.1 and we already have that
version as the minimum required Bazel version, so that we can clean that
up now.
[1] https://github.com/bazelbuild/bazel/issues/5633
[2] https://github.com/bazelbuild/bazel/pull/5647
Change-Id: Ia00e5d7b4eb9c18be808b290ac299e658ab33b9a
|
|
|
|
| |
Change-Id: I78c09f7f0c716ef215904b070c74aa400f9a4c96
|
|
|
|
| |
Change-Id: Ie95734511999c482cd5af453d95e37fd2a490ed8
|
|
|
|
| |
Change-Id: Icb85272a3c1d68700d7b251d643c8640a82adfb7
|
|
|
|
| |
Change-Id: I08b931ab1dce43d52bb066613a0f23f86836a508
|
|
|
|
| |
Change-Id: Ib585fc309183d7edec753d1976ca51d0f9fe29e3
|
|
|
|
| |
Change-Id: Ia2cc08e17d06d81af4d69f3c1e6afb6fff890dfa
|
|
|
|
|
|
|
|
|
| |
Change I9d79b9064 was backported to stable-2.14 and fixes these
warnings, so we no longer need the suppressions.
This reverts commit 89ab8f45874954f6a3b74b29aa8f42cc010f814a.
Change-Id: Ie2e38c12e908809df104e7ebe74921ac9902e42d
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Object#equals is supposed to define an equivalence relation[1][2]. This
includes the property of symmetry: a.equals(b) if and only if
b.equals(a). The old implementation of IdString#equals was asymmetrical:
it returned true when comparing to a String with the same urlencoded
value, but of course String#equals will always return false when
comparing to an IdString.
Fix IdString#equals to only compare equal to other IdStrings. It's
trivial and readable to fix the few usages in Gerrit core.
Since this method is part of the extension API, it's possible (though
personally I think highly unlikely) that plugins in the wild are
depending upon the current behavior, similar to how core was. Breakages
may be subtle, if for example callers were trying to look up Strings in
a Map<IdString, V>. However, it's worth noting that the asymmetric
behavior of equals meant that plugins were _already_ in for subtly
confusing or broken behavior.
At a minimum need to mention this behavior change in release notes.
[1] https://en.wikipedia.org/wiki/Equivalence_relation
[2] Effective Java, Item 8: "Obey the general contract when overriding
equals"
Change-Id: I9d79b90640ea095b71803185d581cda77b93964e
(cherry picked from commit e24bb8be8df1084406d2900c4bd0a7b51bb811d1)
|
|
|
|
| |
Change-Id: I6ababa146a93fcd23481b5f3ce2d55dc4486cdf3
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This release contains the following fixes since 4.7.3.201809090215-r:
- Fix ObjectUploadListener#close
- Fix error handling in FileLfsServlet
- ObjectDownloadListener#onWritePossible: Make code spec compatible
- ObjectDownloadListener: Return from onWritePossible when data is written
- Fix IOException when LockToken#close fails
- Fix NoSuchFileException during directory cleanup in RefDirectory
- Externalize warning message in RefDirectory.delete()
- Suppress warning for trying to delete non-empty directory
Bug: Issue 9667
Change-Id: Icc0a72d4c0301593d13434911e9a41ede4b6071b
|
|
|
|
| |
Change-Id: I9dec94b81079574da0c66bca2860e093f70be8f4
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Chaining the TemporaryFolder with the custom TestRule results in the
directory for the common server being deleted after the first test
method, even though it might still be needed. This means a @UseLocalDisk
class cannot have more than one method, or things break, including but
not limited to the NRT reopen threads throwing exceptions due to the
index files disappearing out from under them. We just got lucky that
none of our local disk tests were actually affected by this.
To make this work, we can't tie the lifetime of the TemporaryFolder to a
single AbstractDaemonTest method invocation. Instead, create the tempdir
within GerritServer#initAndStart, and allow callers to get it back out.
We have to be a bit more careful about managing the lifetime of the
tempdir since some GerritServers, particularly in StandaloneSiteTest,
have a shorter lifetime than their tempdirs. Fortunately, TempFileUtil
make this relatively easy.
As a nice side effect, we can get rid of the hack of passing the tempdir
in via a Gerrit config option.
Change-Id: I80ac66a27bc990f2cef966c9b372c86ce277c471
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Without this change, building fails with:
$ bazel build tools:error_prone_warnings_toolchain
ERROR: /usr/local/google/home/dborowitz/nobackup/c/gerrit-upstream/gerrit/tools/BUILD:45:1: no such target '@bazel_tools//tools/jdk:platformclasspath.jar': target 'platformclasspath.jar' not declared in package 'tools/jdk' (did you mean 'platformclasspath8.jar'?) defined by /usr/local/google/home/dborowitz/.cache/bazel/_bazel_`
This target was renamed in 0.17.1:
https://bazel.googlesource.com/bazel/+/3987300d6651cf0e6e91b395696afac6913a7d66
https://bazel.googlesource.com/bazel/+/35eaec4caae64af9077e7e614edfa62a09c3b0c4/CHANGELOG.md#release-0_17_1-2018_09_14
Change-Id: Ib1ade4041ecf3839e2755c21858882dca9dd30cb
|
|
|
|
|
|
|
| |
New Bazel version is using JDK 9 javac. That will allow us to target
JDK 9 in Gerrit build.
Change-Id: I86666a9fda21c55852c64373bca2131ef0572c72
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The fix applied by I1c88102d4 got rid of the error-prone warning but
missed to address the actual issue in that line of code. The intention
of the original code was to compute a double/float percentage and
mathematically round it to the next integer/long value. However, as
(long * int / long) == long, the computed percentage was a long value,
for which the floor operation was automatically applied. Hence, the
round function was even meaningless. The correct fix is to use a double
value in the operation so that the result is also of type double and
preserves all fractions.
By doing so, we also get rid of the error-prone warning, which only
indirectly hinted at that error.
Change-Id: Iea07a05281faf00b0b52cf25a2fd35ca25b069c5
|
|\ |
|
|/
|
|
|
|
|
|
|
|
|
|
|
| |
The new event is emitted on deletion of a draft change, and deletion of
a regular change.
Plugins may implement the ChangeDeletedListener to perform specific
actions on change deletion.
The event is included in the output of the stream-events ssh command.
Bug: Issue 9711
Change-Id: I37424eafa73703f51072ff8ae6c2546a10ed5c7d
|
|
|
|
| |
Change-Id: Ic68c66d06c615626a6b3561fbd99fb88272a726a
|
|
|
|
|
|
|
| |
DeprecationWarning: This method will be removed in future versions.
Use 'list(elem)' or iteration over elem instead.
Change-Id: Ib4733f8f6200408b06cb014358e5610dfbbc4f11
|
|
|
|
| |
Change-Id: I6cc3b0323731a0ff4aa2d743938742ecc3900c74
|
|
|
|
| |
Change-Id: I79170cbcbc573b037542a439592ca8de063fbd47
|
|
|
|
|
|
|
|
|
| |
single_file attribute is deprecated: [1] and should be replaced with
allow_single_file.
[1] https://docs.bazel.build/versions/master/skylark/lib/attr.html#parameters-3
Change-Id: I72f955ac47764c4eeb1ad65ab1f4b27db590521f
|
|
|
|
|
|
|
|
| |
FileType is deprecated: [1] and should be replaced with list of strings.
[1] https://docs.bazel.build/versions/master/skylark/lib/FileType.html
Change-Id: Ifd9b9ac4ad35ae70aaf7ffc4e358792c80f12486
|
|
|
|
|
|
|
|
|
|
|
| |
PACKAGE_NAME was deprecated in favor of package_name() function and is
going to be removed in future Bazel releases. Moreover Bazel is trying
to set --incompatible_package_name_is_a_function=true per default to
enforce Bazel users to stop using it: [1].
[1] https://github.com/bazelbuild/bazel/issues/5827
Change-Id: I4d1a6ccbeae611fdd0e9c1e9fdd89b6dd1294f36
|
|
|
|
|
|
|
|
|
|
|
| |
This release includes the following fixes since 4.7.2.201807261330-r:
- Fix atomic lock file creation on NFS
- Fix handling of option core.supportsAtomicCreateNewFile
- GC: Avoid logging errors when deleting non-empty folders
- Fix GC run in foreground to not use executor
Change-Id: I6862cd6c3bff5b07234d886f80fb42504dd1ca79
|
|\
| |
| |
| | |
permission" into stable-2.14
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Before this change, the Tags creation form fields were visible also if
either refs/* or refs/head/* was allowed for Create Reference. This fix
limits that visibility to a create refs/tags/* permission solely, as per
current documentation anyway. isOwnerAnyRef() still also makes the panel
visible, overriding potentially missing ref creation permissions.
Create Annotated Tag is still also required for the user to be able to
use the optional Annotation field. In this case, the created tag is no
longer lightweight but becomes annotated. Both kinds of tags are still
supported through such a single Tags creation panel or form, thus the
need to allow both permissions even if aiming for annotated tags only.
(Command line does not have that design limitation indeed.)
Bug: Issue 9689
Change-Id: I7e3d11a62ad1e49575e6ef743138158efa831e6a
|
| |
| |
| |
| | |
Change-Id: I94df398e74616ff9237dde9da4e8d8a88b50ed5e
|
| |
| |
| |
| | |
Change-Id: Ia7925987947465f38be5b5ca4ef7ead970226d57
|
|\ \
| | |
| | |
| | | |
stable-2.14
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
I you have a change URL like:
http://localhost:8080/c/test-project/+/688
Stepping back to parts "../../" will get you:
http://localhost:8080/c/
Wich is _not_ the base url.
Do not try to guess how URL is constructed but trust WebLink providers
to provide an absolute or a relative URL.
Provide correct relative url from Gitweb.
Bug: Issue 5316
Change-Id: I7f404d470a8fe0f32ef94bb2c7b4c6a9ba68a2b8
(cherry picked from commit 69ab4a8569a690da7d6fc0c57e0a74c2d97ef6c0)
|
| | |
| | |
| | |
| | | |
Change-Id: I2fd01cb6775a905d7e602d5415102b274bd98fd3
|
| |/
|/|
| |
| | |
Change-Id: Iafc734d5f3a7e5f4ce9508f93525fe8348dd3b30
|
|/
|
|
|
|
|
|
|
|
|
| |
The call to assertThat(e).isNull() causes ErrorProne to raise a
warning that there is a missing call to fail(), because it thinks
that the exception is expected.
Instead of asserting on the IOException, rethrow it as the cause
of a new ValidationException.
Change-Id: I7dd681196f8d63eb6f2a0063b55aa94d0ec1d71f
|
|
|
|
|
|
|
|
| |
Add grouping parenthesis to make the operator precedence explicit.
See http://errorprone.info/bugpattern/OperatorPrecedence
Change-Id: Ica10d3e1494e9109883ef0c327015e423e91dd31
|
|
|
|
|
|
|
|
| |
Add grouping parenthesis to make the operator precedence explicit.
See http://errorprone.info/bugpattern/OperatorPrecedence
Change-Id: I8a0853f8fac05218e77e1a94ebff989380246d37
|
|
|
|
|
|
|
|
| |
Add grouping parentheses to make the operator precedence explicit.
See http://errorprone.info/bugpattern/OperatorPrecedence
Change-Id: Iaf80a165e640f19bcb95b579cd3aba45ad9b20ba
|
|
|
|
|
|
|
|
|
|
| |
An incorrect target name causes the build to fail with:
no such target '@bazel_tools//tools/jdk:platformclasspath8.jar'
Removing the '8' suffix fixes it.
Change-Id: I915239a5e1ce6403d7b6760d895aa085f00a1b83
|
|
|
|
|
|
|
|
|
|
| |
We don't need to test these versions all the time, so remove the
tests.
Support for versions 6.2 and 6.3 is kept; if we need to check for
regressions we can manually enable the tests again later.
Change-Id: Ia15785ad9cc995b4672814dd346e02a6ee4026a0
|
|
|
|
| |
Change-Id: I4ca4f304cdff6d8275e60856b992ee32d9c61027
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Issue:
Gerrit allows to specify email like username (especially handy
when one wants to maintain unique email addresses across the server).
However as opposite to what is described in [1] it allows only:
._- characters from allowed set that contains !#$%&‘*+–/=?^_`.{|}~
Considering the fact that I was able to perform clone (over SSH
and HTTP) push for review, review and submit with all those
characters but / this patch proposes to extend username allowed
chars to !#$%&‘*+–=?^_`.{|}~ when @ is used (we have email like
username) but keeps original characters set a-zA-Z0-9 when one
uses single character for username.
[1] https://www.mailboxvalidator.com/resources/articles/acceptable-email-address-syntax-rfc/
Change-Id: Iabb82cea5be32a68fe1d9fc65a6ca08743063a3f
Signed-off-by: Jacek Centkowski <jcentkowski@collab.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Change I5f8b333e9 made the receive.maxObjectSizeLimit setting
inherited from the parent project. However this is a change in
behavior that users might not expect, and could have unexpected
results.
Rather than making this new behavior the default, make it so that
it must be explicitly enabled in the site config by setting a new
value, receive.inheritProjectMaxObjectSizeLimit, to true. Default
the value to false when not explicitly set, to maintain backward
compatibility with the behavior before change I5f8b333e9.
Bug: Issue 9528
Change-Id: I63a990a61f0428148a5d26f9cf539d7d419b21ae
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In the current implementation it is possible to set the limit per
project in the project.config on refs/meta/config, and at global
level in $site/etc/gerrit.config. The project setting may override
the global setting if it is lower. Changing the global setting
requires a server restart.
A limitation of this implementation is that we cannot set the limit
at a project level and have it inherited to its child projects; it
is necessary to explicitly set the limit on each child project.
This limitation causes a lot of extra work in the case where for
example we have a project hierarchy like:
|- All-Projects
|
-- Namespace-A
| |
| |-- Project-A
| |-- Project-B
. . ..
. . ..
| |-- Project-X
|
|
-- Namespace-B
Where the Namespace-X projects are assumed to be "parent only"
projects, if we want to set a limit for all the projects under a
namespace hierarchy, we need to set it explicitly on all those
projects individually rather than only on the "Namespace-X".
With this change the limit is inherited from the parent project.
The global limit is still respected, and the project still can't
set a higher value than the global, either explicitly per project
or via inheritance.
Similarly, if no global limit is specified, a child project still
may not set a limit higher than its parent.
The inheritedValue is removed from the config info and replaced
by a summary string describing how the effective value was
inherited or overridden from the parent project or the global
config. This string is used as the tooltip on the effective
value in the UI.
As a side effect of this change, it is now possible to effectively
change the global limit without having to restart the server, by
setting it on the All-Projects project. Note that this only works
if the new limit is lower than what is already configured in the
actual global limit in gerrit.config.
Bug: Issue 9528
Change-Id: I5f8b333e905ed0a147526ae33ff2bab2cbe222ef
|