A Jenkins Pipeline step to convert simple Travis YAML to Pipeline on the fly.
before_install
install
before_script
script
after_failure
after_success
after_script
env
(though not yet excludes/includes or language-specific axes)
Actual guts are all in SimpleTravisRunner.groovy.
An example of using a Jenkinsfile to call a .travis.yml can be found at this repo.
-
config.jelly and(config.jelly not needed for global variables) help HTML for Snippet Generator and reference docs. - Tests, tests, tests! (More tests would probably be good-to-have, but what's there covers existing functionality.)
- Ideally, find a way to optionally output as Pipeline code rather than execute, for migration purposes.
- Make sure this will fail if run outside of a
node
block. - Allow execution of all
script
entries even if one fails. - Time out individual
script
entries. - Determine what language-specific environment axes will be supported natively (i.e.,
rvm
,php
, etc) - Implement environment axes equivalent to Jenkins Matrix jobs, serially initially.
- Figure out how to extrapolate
node
labels for axes to enableparallel
usage. (switched this up - requiringnode
context now and/or specification of a label) - Travis
matrix
inclusions and exclusions from environment axes (andallow_failures
). - (stretch goal) Implement auto-generation of
script
and friends for (some) languages? Debatable whether to do this. - Decide whether to keep emulating Travis's behavior of ignore
after_*
steps' failures when setting build status.