| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Prevent giving review scores but allow commenting for a change in
CI states (STAGING, STAGED or INTEGRATING). No need to block it with
error message like before. Also if the state has changed to one of
CI states while reviewing, review scores are dropped, just comments
are saved.
Task-number: QTQAINFRA-858
Change-Id: I684f004a1d680b2db184fedb50d936337f603721
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Review scores (aka submit records) where not fetched because the
function doing this (getSubmitRecords) utilizes 'canSubmit' function
that returns error for a change in STAGING or STAGED state. Added new
parameter (allowStaged) to function interface to allow bypassing this
check and get review scores fetched.
Task-number: QTQAINFRA-859
Change-Id: I48c06e6a5833e762529efb305ef93182d4202572
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Allowed unstaging also from STAGING state in database update function.
Without this, state was not changed at all even though "Unstaged" change
message was posted. Because change was not moved to NEW state as expected,
this also caused an error message complaining about incorrect state.
Task-number: QTQAINFRA-857
Change-Id: Iea359597584991fb6da17a72cf0c400970cafda6
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Permanently hide the Rebase button so rebasing a change is never
possible from the web UI. Set following value in 'gerrit.config'
to make this happen:
[gerrit]
hiderebase = true
If this value is not set (or is false) the default Gerrit way is on.
Then the change owner and submitters can always rebase changes in the
web UI (even without having the Rebase access right assigned).
Task-number: QTQAINFRA-848
Change-Id: I6e26f6ab5e88d0446d6bcb5081e74976f16cc5d0
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Support for new change state "deferred" to be able to discern
"really dead" from "postponed" changes. This is equal to "abandoned"
for all practical purposes except presentation and filtering. State
transitions are possible between "review in progress", "abandoned"
and "deferred". And menu bar has new "All/deferred" and "My/Deferred
Changes" default filters.
Task-number: QTQAINFRA-598
Change-Id: Ib1801f35eabf6b3b32bf41b0f3b5b496fbcda72a
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
| |
Sanity review category box now starts in hidden mode. All
review categories are now expandable boxes.
Task-number: QTQAINFRA-355
Originally-by: Ola Bjuremo <ola.bjuremo@tieto.com>
Change-Id: I28ba61d00aec52da95d2ae6f890d18ed13fce2ac
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In order to have more precise control of when
e-mail notifications are sent, a filter expression
can now be configured in site configuration.
An example filter configuration for Qt sanity
review bot:
[review]
filter = SRVW < 0 || reviewer != qt_sanity_bot
If filter evaluates true the e-mail will be sent
Task-number: QTQAINFRA-220
Originally-by: Tomas Ljunggren <tomas.ljunggren@tieto.com>
Change-Id: Ia23301f49397f017684ada7f7653781fd8770108
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Remove leading and trailing whitespace in username when creating a request
to avoid incorrect login. Valid username and extra whitespace with valid
password causes creation of new Gerrit user that logs in. This is unwanted
behavior and because whitespace is invalid character to use in username it
can be safely removed.
Task-number: QTQAINFRA-456
Change-Id: I0315285a2cc5b349515ba57c0d96167044238996
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
| |
Patch http://forge.typo3.org/attachments/download/17860/http-anon-access.patch
applied in order to allow anonymous access to gerrit.
Task-number: QTQAINFRA-357
Change-Id: I830b8086329248c564a476150064fe14acf65477
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When using cherry pick submit type, there is now additional options
which define which footers are added to the commit message. Also,
added footers can be limited to reviewers who gave the maximum
approval score. These options are visible only for actual projects,
not for All-Projects.
Bug: issue 655
Originally-by: Mika Hamalainen <mika.hamalainen@accenture.com>
Change-Id: Ibfb7193edde7f88607683adf061e8603cb382d06
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
| |
Added a button to stage a change directly from one page review.
Change-Id: I7d124edc611c292c22664809702c53e44ee90ddf
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Staging branch allows user to move changes to staging branch for further
testing before they are merged into repository. Staging branches are located
in folder refs/staging/<branch>. <branch> matches branch under refs/heads.
A REST style API is used to stage/unstage a patch:
To merge a patch to staging:
/changes/{id}/revisions/{sha1}/stage
To remove a patch from staging:
/changes/{id}/revisions/{sha1}/unstage
A staging branch is moved to testable build with SSH commands. Build refs
are under refs/builds.
Staging branch is updated in the following situations
- User moves a new change to staging. The change is merged to staging branch.
- User creates a new build. Staging is re-created from refs/heads.
- User merges a build to refs/heads. The staging branch is updated to match
the updated head.
Changes include: new button in change screen to stage/unstage (triggering REST
API call), new category called 'staging', SSH commands for managing staging and
build branches.
Change-Id: I4334f50f0bf74606bbbd6bff553fee6e4f019b49
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
| |
Replaced the existing Gerrit favicon with Qt's favicon.
Change-Id: I3850ab834d9678e08f94f836c6702de924ab7f25
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
When registering a new email address with gerrit, it sends an email
with a verification link. This link now uses 'https:' schema instead
of 'http:'.
Task-number: QTQAINFRA-167
Originally-by: Peter Heubeck <peter.heubeck@tieto.com>
Change-Id: I0e2e5a88171d6d6519149f5a2c2248810436c156
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
| |
Second and additional changes in the same submit attempt must be built
on top of the previously cherry picked change, not the branch tip.
Change-Id: I9183f68b5780af1f28cb357217663e4bdb2d4e76
(cherry picked from commit cd60499cd65812d926806e49d15a6852c7798871)
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
This would at least allow the partial changes to the
database to rollback and not get users in a bad state when
e.g. in the case of OrmDuplicateKeyException: patch_set_ancestors'.
Bug: issue 2034
Bug: issue 2246
Change-Id: I65e111bedc3cec033299b5b5360aaee733d4b5c6
(cherry picked from commit eb4778d079e7156c82d04c2f26c71ca89d12df16)
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
| |
Added choices for review labels and button to publish review comments
from one page review.
Change-Id: I2841a80d715fb44c2e157e3eb44e381d17ab2492
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A complete change can be reviewed from single page. The new screen is
launched from standard 'All Side-by-Side' and 'All Unified' links on the
change screen. One configuration change is needed in 'gerrit.config' to
make this happen:
[gerrit]
onepagereview = true
If this value is not set (or is false) the default Gerrit way kicks in
and every file in change is opened in separate tab.
Originally-by: Mika Hamalainen <mika.hamalainen@accenture.com>
Change-Id: Ia075fdec58a86dec2534381e00fafb75e48d3d78
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Ismo Haataja <ismo.haataja@digia.com>
|
|
|
|
| |
Change-Id: I49a7fc97fd6d4f199c25774faa50ec4a826ae64a
|
|\
| |
| |
| |
| |
| |
| | |
* stable-2.6:
Fix installation of plugins
Change-Id: I6742a5cd3417633be2ad8347f8ebea2fdcc07b88
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
e51c428fe5ef3960f4990e1c0471d067783d4718 broke the unregistering of plugin
owned SSH command by passing wrong instance to the ConcurrentMap.remove()
function.
bae9e58ddd68992473307787fb01e022f5969fee was trying to fix it, but broke it
for the installation of plugins by throwing the `IllegalStateException` in case
the old registration handle wasn't found. But it's perfectly possible and even
wanted during plugin reloading: registration handle is replaced and the old one
cannot be found.
Moreover, bae9e58ddd68992473307787fb01e022f5969fee didn't fix all places:
the same bug in unregistering of SSH commands in
DispatchCommandProvider.replace() method wasn't fixed.
Change-Id: I21132d4f7aec58347acbf50d75a962cb8251201f
|
|\ \ |
|
| |\|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
* stable-2.6:
Remove documentation links from admin page
NPE when deleting draft patch set when previous draft already deleted
Fix example for 'test-submit rule' in Prolog cookbook
Change-Id: Ib96358c1fe6a953d17e88fd9f5bd4832fd51334a
|
| | |\
| | | |
| | | |
| | | | |
deleted" into stable-2.6
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
If a change has 3 patch sets, all of which are drafts, and draft
patch set 2 is deleted, attempting to delete draft patch set 3 (the
current patch set) will result in NPE.
The reason is that on deleting a draft patch set which is the current
patch set, the current patch set is set to the previous patch set
whose ID is calculated by subtracting 1 from the deleted patch set ID.
This does not work if the previous patch set has already been deleted.
Instead, set the current patch set to the highest patch set ID that
is available.
Bug: Issue 2128
Change-Id: I235d4d943a5525e8223218a97f4702e26f09a69e
|
| | |/
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The links for Change-Id and Signed-off-by documentation in the
admin page are hard-coded to specific versions on the the
gerrit-documentation.googlesource.com site.
This causes two problems. Firstly, they are not links to local
versions, and secondly it is easy to forget to update them and
they become out of date.
Remove the links. Administrators should be capable of finding
the relevant pages in the documentation by themselves.
Bug: Issue 2127
Change-Id: Idd4ddd1126e0631615d564173a02e3047aa87559
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Since commit b1f730b89458d9d094bc93f3d328ad744c39bbb5 it is
'test-submit rule' and not anymore 'test-submit-rule'.
Change-Id: I4c32893afa1410af5347a8b659c5811aac218f63
Signed-off-by: Edwin Kempin <edwin.kempin@sap.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This behavior is caused by submitting a commit that has a tag.
Method MergeUtil.markCleanMerges is used by each merge strategy
to update the status of a change if it was a clean merge. It skips
commits that are identified as accepted and their entire ancestry
chain. If a commit is tagged, method MergeOp.getAlreadyAccepted
identifies it as accepted (even if it is not merged yet).
Fix prevents a commit that is referred by a tag to be included in
alreadyAccepted set. If such commit is already merged then it will
be covered anyway by adding heads to alreadyAccepted.
Add a corresponding test that pushes a commit with a tag
Bug: Issue 600
Change-Id: If00247b809b985eaf60ef5ef09fad0f475fb06b9
(cherry picked from commit b22ee233f1ef880cd84beb5939e332de1d7d704e)
|
|\ \ \
| |/ /
|/| |
| | | |
stable-2.7
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
If a change in the merge queue is marked ALREADY_MERGED, it is likely a
result of a bug in the merge code. It means the change has been merged,
but neither properly marked as such in the DB nor propagated to hooks for
sending email notifications or being mentioned in stream-events.
Calling setMerged() instead of setMergedPatchSet() properly marks the
change as merged in the database and runs hooks for sending emails and
events.
This is an improved version of I44548cce039e303a86d579b0b0d76b12ad9bdcfb
and fix for issue 2044.
Bug: issue 2044
Change-Id: I4e95a74e58ff51a54ed54a02298c719e19f6fd0d
|
|\ \ \
| | |/
| |/|
| | |
| | | |
* stable-2.6:
Fix reenabling of plugins
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
e51c428fe5ef3960f4990e1c0471d067783d4718 broke the unregistering of plugin
owned SSH command by passing wrong instance to the ConcurrentMap.remove()
function. Currently plugins can not be reenabled.
Change-Id: I2b2dfe6ffe89a426a03d945d9695bc35e7fe9f1a
|
|\| |
| |/
|/|
| |
| |
| |
| |
| | |
* stable-2.6:
Fix: RepoCommandLink's text is wrong
fix: NoOp score not displayed for open changes
Change-Id: Ib7ba9c2030740422cbda5d95964ab6aab1e96cf2
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The title of the "repo download" link was mistakenly set to "checkout"
during refactoring in commit cbc629f4f4ce6bd9071a540cb026ef8a9d53382c.
Change it to the correct "repo download" title.
Bug: Issue 2093
Change-Id: I4c30d0fedf56942f741edbbfaa7505052e5e370c
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Score for NoBlock/NoOp category doesn't get displayed for open changes
on All --> Open or My --> Changes. For such category, none of the label
flags i.e. recommended, liked, approved or rejected flags are set for
open changes.
Further, abandoned and merged change display different images for same
score e.g. If a -1 has been assigned, Abandoned change displays a "X"
whereas a Merged change displays "-1". A Merged can be rejected for a
NoBlock or NoOp category type.
Bug: Issue 2056
Change-Id: I29cfa02f4a43ef6be1cb94614ed75b979a84f7af
(cherry picked from commit e88304abce3ceb9c7b93040363c4db3219459b0a)
|
|\|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
* stable-2.6:
Fix browser NPE when ChangeCache is incomplete
Conflicts:
gerrit-gwtui/src/main/java/com/google/gerrit/client/changes/ChangeDescriptionBlock.java
gerrit-gwtui/src/main/java/com/google/gerrit/client/changes/PublishCommentScreen.java
Change-Id: Ib1c1166426c65c5ae44cb2146bc4566b38c16cc4
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The ChangeCache is buggy and does not always populate its
contents. Unfortunately it is also designed to be unable to
load missing contents on demand, causing consumers to NPE.
ConfigInfoCache is a better design for this sort of lazy
loading and reuse of data.
ChangeCache can be missing information if:
1) opens side-by-side view in a new tab;
2) user presses 'r' to open publish comment screen
Instead of using the unreliable ChangeCache, stub out a
ChangeDetail to feed to the info block on the publish
comments screen.
Bug: issue 2039
Change-Id: Id542528d93af1cab49b001ca5e90addfc5d05b78
(cherry picked from commit f646aa77fb67b73c8724d7a632cafd441365e901)
|
|\|
| |
| |
| |
| | |
* stable-2.6:
Expand capabilities of ldap.groupMemberPattern
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Previously, the pattern supported a limited set of variables that were
either accidentially available (due to their use in other queries) or
hard-coded (e.g., `username` is a special-case that was added).
Furthermore, the documentation made reference to being able to use
variables such as `${uidNumber}` even though they are not actually
supported (since `uidNumber` is normally never queried). Under the
default RFC 2307 configuration of LDAP, the only variables available
were `displayName, `mail`, `uid`, and `username` (It's noteworthy
that `username` was added as a special-case due to the default
`groupMemberPattern` containing `${username}` even though `username`
is substitued by Gerrit and not LDAP).
This changeset removes the artificial restrictions on the attributes
used in the `groupMemberPattern`. Any variable is assumed to
originate from the account, but `username` is still overridden and
provided by Gerrit (as before). This allows more expressive patterns,
which allows us to fix an outstanding bug in group matching. Prevously,
a user whose `gidNumber` matched the group's `gidNumber` would not have
been included in the group. This changeset updates the default
`groupMemberPattern` to account for this issue by adding the additional
case of `(gidNumber=${gidNumber}`.
Bug: Issue 2054
Change-Id: Iff3a14c569a10c1ef693b672f4710fb6f2f8d9a6
|
|\|
| |
| |
| |
| | |
* stable-2.6:
Fix Gerrit plugins under Tomcat by avoiding Guice static filter
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The static GuiceFilter breaks when a plugin loads and configures
its own web environment. Follow the pattern used by JettyServer
to initialize and load GuiceFilter in a non-static way, allowing
Gerrit to always have its own filter stack during any request.
Bug: issue 1966
Change-Id: I864f3badc4ec90154a276d35e6a5a3fa75058933
|
|\|
| |
| |
| |
| |
| |
| | |
* stable-2.6:
Say in gsql documentation that 'Access Database' capability is required
Release notes for Gerrit 2.6.2
Fix NullPointerException when comparing AccessSections
|
| |
| |
| |
| |
| |
| |
| |
| | |
Since abd6d4e1d4ac11b7478c00289258a462d8544a24 the 'Access Database'
capability is required to run the gsql command.
Change-Id: I39b2b22be975172839ab9808649afcfc58994806
Signed-off-by: Edwin Kempin <edwin.kempin@sap.com>
|
| |
| |
| |
| | |
Change-Id: I601b56793f265603d608bc397f386d35c0cdf3b8
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
AccessSection.permissions and Permission.rules may not be set and null
values should be handled in the equals methods.
We saw this NPE when PermissionCollection tried to sort the access
sections.
Change-Id: I6f5a0886e6f9e5498fe9d756f289c5e91ed3485e
Signed-off-by: Edwin Kempin <edwin.kempin@sap.com>
(cherry picked from commit 915104def3969fe5153bdc1e3e22f007483ff1aa)
|
| |
| |
| |
| |
| |
| | |
terms have changed in 2.6. schema still hasn't.
Change-Id: I1003d420fcd5b2f56b9c1dc7955f761c37253dc5
|
|\ \ |
|
| | |
| | |
| | |
| | | |
Change-Id: I35dadcce6e151ccbbabe4d443262ff8e2b457f3e
|
|\| | |
|
| | |
| | |
| | |
| | | |
Change-Id: I99c8979725ced1a6191d4a5f94ab70378c5b260d
|