luke
/
sbt
mirror of https://github.com/sbt/sbt.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
sbt/.travis.yml

57 lines
2.6 KiB
YAML
Raw Normal View History

Switch to xenial image The majority of Travis CI builds have probalby migrated to using Xenial now, so it's better to migrate with the pack.
2019-10-03 22:40:51 +02:00
dist: xenial
Add cache setting to Travis
2015-02-17 18:02:41 +01:00
First attempt at travis-ci integration for sbt.
2014-02-21 02:11:17 +01:00
language: scala
Use the new Docker queue on Travis faster, better, stronger vms boot in seconds, network is more stable also cleaned up the .travis.yml file a little :)
2014-08-22 09:53:26 +02:00
env:
Add, configure & enforce sbt-whitesource
2018-01-11 15:13:21 +01:00
global:
test everything on JDK11
2019-01-04 04:32:32 +01:00
- TRAVIS_JDK=adopt@1.11.0-1
Upgrade scala version to 2.12.12 This introduces `Def.unit(...)` to workaround the pure expression does nothing warning.
2020-08-12 03:06:57 +02:00
- SCALA_212=2.12.12
Add util tests to the jdk8 build The existing .travis.yml in the util project runs: mimaReportBinaryIssues scalafmtCheckAll whitesourceCheckPolicies test The first three are already handled by the first build in the matrix. I add the util tests to the first jdk11 build and the jdk8 build.
2019-12-08 01:48:37 +01:00
- SCALA_213=2.13.1
- UTIL_TESTS="utilCache/test;utilControl/test;utilInterface/test;utilLogging/test;utilPosition/test;utilRelation/test;utilScripted/test;utilTracking/test"
define whitesourceOnPush whitesourceOnPush calls whitesourceCheckPolicies and whitesourceUpdate on push. Since Travis CI secrets are not available during PR from a fork, there's no point in calling these during the PR validation.
2018-12-14 01:31:56 +01:00
# WHITESOURCE_PASSWORD=
Use latest sbt version in a travis build Moving forward, it would be great for sbt to use the latest code to run the scripted tests. This could help detect regressions that aren't noticed by manual dogfooding. We can do this by first publishing 1.4.0-SNAPSHOT binaries before running tests. I didn't want to do this for all of the builds because it does make each build that publishes binaries take about a minute longer. The last build in the matrix is much faster than the other builds though so I made just that build use the snapshot version. Since that build almost always finishes first, this should have little effect on the actual time to run ci even though the cpu time will increase a bit.
2020-08-19 20:36:23 +02:00
- SBT_VERSION_PROP=""
- SBT_LOCAL=false
Add, configure & enforce sbt-whitesource
2018-01-11 15:13:21 +01:00
- secure: d3bu2KNwsVHwfhbGgO+gmRfDKBJhfICdCJFGWKf2w3Gv86AJZX9nuTYRxz0KtdvEHO5Xw8WTBZLPb2thSJqhw9OCm4J8TBAVqCP0ruUj4+aqBUFy4bVexQ6WKE6nWHs4JPzPk8c6uC1LG3hMuzlC8RGETXtL/n81Ef1u7NjyXjs=
Use the new Docker queue on Travis faster, better, stronger vms boot in seconds, network is more stable also cleaned up the .travis.yml file a little :)
2014-08-22 09:53:26 +02:00
matrix:
Check server test project scalafmt on travis
2020-08-05 02:08:30 +02:00
- SBT_CMD="mimaReportBinaryIssues ; javafmtCheck ; Test / javafmtCheck; scalafmtCheckAll ; scalafmtSbtCheck; serverTestProj/scalafmtCheckAll; headerCheck ;test:headerCheck ;whitesourceOnPush ;test:compile; publishLocal; test; serverTestProj/test; doc; $UTIL_TESTS; ++$SCALA_213; $UTIL_TESTS"
Consolidate travis builds There is a large fixed overhead of about O(7minutes) per travis run. We seem to be limited to five build agents at a time so I consolidated the .travis.yml into five builds. I tried to distribute the tasks fairly evenly so that all of the builds take about the same amount of time. On my personal travis build, where I only get three build agents, the total run time of the build dropped to 50 minutes from one hour forty minutes.
2019-05-28 21:07:19 +02:00
- SBT_CMD="scripted actions/* apiinfo/* compiler-project/* ivy-deps-management/* reporter/* tests/* watch/* classloader-cache/* package/*"
Split to MiniDependencyTreePlugin Ref https://github.com/sbt/sbt/pull/5880 This split the dependency-graph plugin into MiniDependencyTreePlugin and DependencyTreePlugin.
2020-09-23 03:21:01 +02:00
- SBT_CMD="dependencyTreeProj/publishLocal; scripted dependency-graph/* dependency-management/* plugins/* project-load/* java/* run/* nio/*"
Consolidate travis builds There is a large fixed overhead of about O(7minutes) per travis run. We seem to be limited to five build agents at a time so I consolidated the .travis.yml into five builds. I tried to distribute the tasks fairly evenly so that all of the builds take about the same amount of time. On my personal travis build, where I only get three build agents, the total run time of the build dropped to 50 minutes from one hour forty minutes.
2019-05-28 21:07:19 +02:00
- SBT_CMD="repoOverrideTest:scripted dependency-management/*; scripted source-dependencies/* project/*"
Use the new Docker queue on Travis faster, better, stronger vms boot in seconds, network is more stable also cleaned up the .travis.yml file a little :)
2014-08-22 09:53:26 +02:00
use AdoptOpenJDK 8
2018-11-28 00:48:22 +01:00
matrix:
fast_finish: true
include:
add some AdoptOpenJDK JDK11 tests
2018-12-14 21:36:16 +01:00
- env:
Use latest sbt version in a travis build Moving forward, it would be great for sbt to use the latest code to run the scripted tests. This could help detect regressions that aren't noticed by manual dogfooding. We can do this by first publishing 1.4.0-SNAPSHOT binaries before running tests. I didn't want to do this for all of the builds because it does make each build that publishes binaries take about a minute longer. The last build in the matrix is much faster than the other builds though so I made just that build use the snapshot version. Since that build almost always finishes first, this should have little effect on the actual time to run ci even though the cpu time will increase a bit.
2020-08-19 20:36:23 +02:00
- SBT_LOCAL=true
Bump sbt snapshot version
2020-10-06 07:15:53 +02:00
- SBT_VERSION_PROP=-Dsbt.version=1.4.1-SNAPSHOT
Bump travis java 8 version
2019-09-27 21:20:01 +02:00
- TRAVIS_JDK=adopt@1.8.0-222
Add util tests to the jdk8 build The existing .travis.yml in the util project runs: mimaReportBinaryIssues scalafmtCheckAll whitesourceCheckPolicies test The first three are already handled by the first build in the matrix. I add the util tests to the first jdk11 build and the jdk8 build.
2019-12-08 01:48:37 +01:00
- SBT_CMD="++$SCALA_213; $UTIL_TESTS; ++$SCALA_212; $UTIL_TESTS; scripted actions/* source-dependencies/*1of3 dependency-management/*1of4 java/*"
use AdoptOpenJDK 8
2018-11-28 00:48:22 +01:00
before_install:
- curl -sL https://raw.githubusercontent.com/shyiko/jabba/0.11.0/install.sh | bash && . ~/.jabba/jabba.sh
Use latest sbt version in a travis build Moving forward, it would be great for sbt to use the latest code to run the scripted tests. This could help detect regressions that aren't noticed by manual dogfooding. We can do this by first publishing 1.4.0-SNAPSHOT binaries before running tests. I didn't want to do this for all of the builds because it does make each build that publishes binaries take about a minute longer. The last build in the matrix is much faster than the other builds though so I made just that build use the snapshot version. Since that build almost always finishes first, this should have little effect on the actual time to run ci even though the cpu time will increase a bit.
2020-08-19 20:36:23 +02:00
- if [ $SBT_LOCAL == true ]; then sbt -Dsbt.io.virtual=false publishLocalBin; fi
Remove SNAPSHOTS from sbt boot directory If the sbt snapshot builds are cached, then we don't actually dogfood sbt when we publish a local sbt version because sbt will load with the version from the last build
2020-09-27 18:45:24 +02:00
- rm -r $(find $HOME/.sbt/boot -name "*-SNAPSHOT") || true
use AdoptOpenJDK 8
2018-11-28 00:48:22 +01:00
install:
- $JABBA_HOME/bin/jabba install $TRAVIS_JDK
Consolidate travis builds There is a large fixed overhead of about O(7minutes) per travis run. We seem to be limited to five build agents at a time so I consolidated the .travis.yml into five builds. I tried to distribute the tasks fairly evenly so that all of the builds take about the same amount of time. On my personal travis build, where I only get three build agents, the total run time of the build dropped to 50 minutes from one hour forty minutes.
2019-05-28 21:07:19 +02:00
- $JABBA_HOME/bin/jabba install openjdk@1.10
Split server unit test on its own Ref https://github.com/sbt/sbt/issues/4186 This is an attempt to fix the flaky server test issue.
2018-06-18 10:20:50 +02:00
- unset _JAVA_OPTIONS
use AdoptOpenJDK 8
2018-11-28 00:48:22 +01:00
- export JAVA_HOME="$JABBA_HOME/jdk/$TRAVIS_JDK" && export PATH="$JAVA_HOME/bin:$PATH" && java -Xmx32m -version
Undo _JAVA_OPTIONS enrivonment variable Ref sbt/sbt#3520 This undoes the environment variable addedin 2017-09-06 update of Travis CI: https://docs.travis-ci.com/user/build-environment-updates/2017-09-06/
2017-09-13 03:29:40 +02:00
Add cache setting to Travis
2015-02-17 18:02:41 +01:00
script:
Try another travis structure + build tweaks I've tried to put together some scripted tests to remove the overhead of compiling the whole sbt, which is around 3 minutes every time. This new structure *should* make the scripted tests run faster. Aside from this, we do some more tweaks: * Increase memory. * Fork processes to compile and run (to see if it makes a difference). * Pass in the server flag to sbt.
2017-05-03 00:59:51 +02:00
# It doesn't need that much memory because compile and run are forked
Use latest sbt version in a travis build Moving forward, it would be great for sbt to use the latest code to run the scripted tests. This could help detect regressions that aren't noticed by manual dogfooding. We can do this by first publishing 1.4.0-SNAPSHOT binaries before running tests. I didn't want to do this for all of the builds because it does make each build that publishes binaries take about a minute longer. The last build in the matrix is much faster than the other builds though so I made just that build use the snapshot version. Since that build almost always finishes first, this should have little effect on the actual time to run ci even though the cpu time will increase a bit.
2020-08-19 20:36:23 +02:00
- sbt -Dsbt.io.virtual=false $SBT_VERSION_PROP -Dsbt.ci=true -J-XX:ReservedCodeCacheSize=128m -J-Xmx800M -J-Xms800M -J-server "$SBT_CMD"
bumping the file to run the build again fixed an issue with the Docker workers
2014-08-22 11:41:21 +02:00
Check formatting with TravisCI Fixes #2657
2016-07-15 00:34:09 +02:00
before_cache:
Add coursier cache to CI These locations are taken from https://get-coursier.io/docs/cache.
2019-08-01 17:57:14 +02:00
- find $HOME/.cache/coursier/v1 -name "ivydata-*.properties" -delete
- find $HOME/.ivy2 -name "ivydata-*.properties" -delete
- find $HOME/.sbt -name "*.lock" -delete
Remove SNAPSHOTS from sbt boot directory If the sbt snapshot builds are cached, then we don't actually dogfood sbt when we publish a local sbt version because sbt will load with the version from the last build
2020-09-27 18:45:24 +02:00
- rm -r $(find $HOME/.sbt/boot -name "*-SNAPSHOT") || true
use AdoptOpenJDK 8
2018-11-28 00:48:22 +01:00
cache:
directories:
Add coursier cache to CI These locations are taken from https://get-coursier.io/docs/cache.
2019-08-01 17:57:14 +02:00
- $HOME/.cache/coursier/v1
use AdoptOpenJDK 8
2018-11-28 00:48:22 +01:00
- $HOME/.ivy2/cache
- $HOME/.sbt/boot
- $HOME/.jabba