Try server, usually just referred to as try, is the easiest way to test a change without actually checking anything into a core repository. The change will undergo the same kinds of builds and tests as if it had landed on a regular integration branch, but will not get merged with mozilla-central.
Try is just another mercurial repository (like inbound, autoland or mozilla-central) with a few key differences:
- Pushing new heads is allowed.
- It is non-publishing.
The first point means that you’ll never need to pull and rebase before pushing, each push creates a new head. In fact, the ability to push from old changesets is a very valuable property of try which is often used for things like regression hunting.
The second point means that draft changesets (changesets that only exist in your local repository), will remain in the draft state after pushing. Normally when pushing to an integration branch, a changeset gets marked ‘public’. This ensures changes that are shared with others don’t accidentally get mutated. Pushing to try doesn’t actually share the changeset with anyone, so changesets remain in the ‘draft’ state and they are still ok to mutate.
Before you can push to try, you’ll need to have the proper credentials and do some light setup. See the Configuring Try page for more information.
The recommended way to push to try is via the
mach try command. This will work with mercurial
push-to-try extension) and git (via
mach try command offers
a variety of different
selectors which are implemented as a subcommands. See selectors for available list.
If no subcommand is specified,
mach try will a subcommand to dispatch to. By default this is
syntax selector. In other words, these commands are equivalent:
$ mach try $ mach try syntax
You can choose to use a different default selector by configuring your
Attaching new jobs from a review
For every patch submitted for review in Phabricator, a new Try run is automatically created.
A link called
Treeherder Jobs can be found in the
Diff Detail section of the review in
This run is created for static analysis, linting and other tasks. Attaching new jobs to the run is
easy and doesn’t require more actions from the developer.
Indices and tables