Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
BitbucketBuildListener was not invoked on
pipeline jobs
Change BitbucketBuildListener to extend from
RunListener<Run<?, ?>> so it will invoked to
update job description and PR build result
|
|
looks like pipeline jobs don't have a disable option
|
|
Replace use of AbstractProject with Job to fix
ClassCastException thrown when using the plugin
in a pipeline job
The fix is "not" complete, it remains to find a
replacement to the call AbstractProject.isDisabled
used to decide if the build should be skipped
Previously called in BitbucketBuildTrigger.run()
fixes #94
|
|
Refactor to move default value for commentTrigger to config.jelly. Handle special case in BitbucketBuildRepository where trigger field is null within unit tests (since BitbucketBuildTrigger can't be mocked or initialized in the test).
|
|
|
|
Defaults to prior "test this please", can be overridden in the job
configuration.
|
|
Add option to cancel outdated builds when a pull request is updated
|
|
Remove all "e.printStackTrace()" calls
|
|
On the latest version I started getting these errors in the logs and my jobs would trigger every minute and never flag the build as having run before.
```
POST state INPROGRESS to https://bitbucket.org/api/2.0/repositories/null/null/commit/b6f8169ce460/statuses/build with key 515f53a20bf37bd484f2be5cda6ab168cd194cee with response {"type": "error", "error": {"message": "Repository null/null not found"}}
```
Functionality in `getOwnerName` and `getRepositoryName` name changed from previous versions, moved to inside `setFullName` which is now doing the `if` on the wrong value. This causes `null/null` to be set as the owner name / repo name.
|
|
|
|
|
|
Use the 2.0 Bitbucket pull request comment API.
|
|
Set an explicit HTTP request timeout
|
|
|
|
|
|
The V1 pull request comment API is deprecated. Replace with a call
to the V2 API. The V2 API is paginated and may need to be called
multiple times to get the entire set of comments.
While fixing this I noticed that while fetching pull requests the
plugin is using the v2 API there, which is also paginated. So not
all open pull requests were being retrieved.
|
|
actually look all on the same line separated by some *. It's incomprehensible to understand how use this property. Only seeing sources now I have idea what the field "branch filter" does.
I fix the html so that will look like as designed by author.
|
|
Fix TTP only working for one of multiple jobs
|
|
move it back to BitbucketBuildTrigger to fix the
the problem that a lot of empty "Parameters"
links are added to each build
|
|
Use EnvironmentContributor instead of build
parameters
Based on https://github.com/jenkinsci/ghprb-plugin/pull/344
Fixes: #86
|
|
The bug was that it did not handle TTBuild
comments from another job
Will now check all comments if they is a TTBuild
for this job, and only trigger the job if we
did not find a match
Fixes: #82
|
|
Use JenkinsLocationConfiguration to fetch the URL.
|
|
`Jenkins.getInstance().getRootUrl()` returns `null` despite the URL
being set through the web UI. `JenkinsLocationConfiguration.getUrl()`
returns the correct URL.
Thanks Esteban Angee on StackOverflow:
http://stackoverflow.com/questions/15949946/programatically-access-jenkins-url
[Resolves #19]
|
|
Env variable "pullRequestAuthor" contain two parts of author: full name of user and short username in format, like: "Some Doo <@somedoo>".
Add test for PR build filter: now we can filter also by author: "a:maxvodo" by example.
|
|
|
|
Using HttpMethodBase.releaseConnection() for this stuff.
|
|
Constructor of ApiClient now using just <T extends HttpClientFactory> T type to make possible using of HttpClient interceptor in tests.
|
|
Using by default only username and passwords fields to provide BitBucket auth cause NullReferenceException in BitbucketRepository::getCredentials(),
because credentialsId is null. It's my fault :(
|
|
Refactoring method ApiClient::computeAPIKey() to check maximum size of computed key.
Now field size contolled by special constant ApiClient::MAX_KEY_SIZE_BB_API.
This changes will resolve, in my opinion, issue with continuously rebuilding: method ApiClient::hasBuildStatus() fail and PR's are always trigger new builds.
|
|
By default we using username/password from original text fields (config.jelly).
If credentials ID provided in plugin settings - using them for BitBucket API auth.
|
|
https://github.com/kadaan/bitbucket-pullrequest-builder-plugin into CredentialsSupport
Conflicts:
.gitignore
src/main/java/bitbucketpullrequestbuilder/bitbucketpullrequestbuilder/BitbucketRepository.java
|
|
If available TTP (aka "test this please") comment Jenkins PR builder continuously rebuilding PR. Now Jenkins post specific build comment.
If you want to rebuild already rebuilded PR, post new TTP comment.
|
|
Now we can using more complex filters with source/destination filtration. Also add experimental integration with Git SCM plugin.
|
|
|
|
|
|
# Conflicts:
# src/main/java/bitbucketpullrequestbuilder/bitbucketpullrequestbuilder/BitbucketRepository.java
|
|
News: http://blog.bitbucket.org/2015/11/18/introducing-the-build-status-api-for-bitbucket-cloud/
API: https://confluence.atlassian.com/bitbucket/buildstatus-resource-779295267.html
|
|
|
|
when configuring BitbucketBuildTrigger and change BitbucketRepository to use the credentials when setting up the ApiClient.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Specify git revision
|
|
|