We can now define a list of "tags" for a task. Specifying "-j <tag>"
in Try syntax will run all tasks having that tag.
MozReview-Commit-ID: Ih9Z0tRZ5VA
This uses a repacked version of the upstream 1.8 nightly
builds. We need 1.8 for WinXP support, but it's safe to
use nightly until this stablises because other platforms
are using stable rust and will catch any nightly-only
feature use.
For these builds I used the channel-rustc-nightly manifest instead
of the older channel-rust-* manifest, to get the separate std packages.
tar xf rustc-nightly-i686-pc-windows-msvc.tar.gz
rustc-nightly-i686-pc-windows-msvc/install.sh --prefix=rustc/
tar xf rust-std-nightly-i686-pc-windows-msvc.tar.gz
rust-std-nightly-i686-pc-windows-msvc/install.sh --prefix=rustc/
tar cjf rustc.tar.bz2 rustc/*
Firefox's automation currently tends to run all the jobs all the time.
It is wasteful to do this. For example, running ESLint when the commit
only changes a .cpp file adds no value.
This commit adds support for only running tasks when certain files
change. The new-style tasks introduced by the previous commit have been
taught a "when" dictionary property that defines conditions that should
hold for the task to be executed. We define a "file_patterns" list that
defines lists of mozpack path matching expressions that will be matched
against the set of files changed by the changesets relevant to the
changeset being built. The eslint task has been updated to only run if
files related to it change.
Because conditions may not be accurate, we add a CLI argument to ignore
conditions and force all would-be-filtered tasks to run.
MozReview-Commit-ID: 3OeBSKAQAeg
Currently, tasks are either "build" or "test" tasks. And "test" tasks
are dependent on "build" tasks, so they are effectively an extension of
"build" tasks.
Not everything is a "build" task. Not everything is associated with a
specific platform.
This commit introduces support for defining non-build "tasks" under the
"tasks" top-level element of a jobs YAML file. Interally, they are
treated as "build" tasks but are declared differently.
By default, all these tasks run.
The -j/--job argument has been added to the try syntax parser. It
specifies an opt-in list of these non-build tasks to run. By default, it
runs all of them.
The eslint-gecko "build" task has been moved to this new mechanism.
Documentation for the new task type have been added.
There is definitely some wonkiness in this implementation. For example,
there are references to "build_name," "build_type," and "build_product,"
which arguably are no longer relevant to generic tasks. However, they
appear to be so integrated into task processing (including route names)
that I'm a bit scared to change them.
MozReview-Commit-ID: BY219tLFb6Z
It is possible to hook up in-tree documentation to Sphinx. Convert the
one-off README.md to ReStructuredText and add it to the Sphinx docs.
I added a moz.build file under testing/ because I don't think it is
appropriate for the Sphinx directive to live in the root moz.build file.
MozReview-Commit-ID: 90tCb7mA63C
We're about to introduce a mechanism to influence which tasks run based
on what files change. To help debug what's happening, print out the list
of commits that influence the task selection.
MozReview-Commit-ID: Kfj2pf1PSIS
Over in bug 1247802 we deployed a new JSON web API on hg.mozilla.org
that returns JSON metadata for changesets that are relevant for build
automation. It returns a superset of what is returned by the pushlog
JSON API. So we switch to it.
MozReview-Commit-ID: 6X3NANo1mgq
Before, we attempted to build and query a URL that potentially had
"None" in it. This printed some wonky messages in the log and may have
contributed to added latency due to the HTTP request that was doomed to
fail.
MozReview-Commit-ID: JrR5PK33vCn
requests should *always* be used for performing HTTP requests because it
has a better API *and* has sane security defaults compared to the HTTP
request APIs in the Python standard library. Although, Python 2.7.9+
does have slightly saner defaults in the standard library. I still trust
requests more.
MozReview-Commit-ID: GqohpfYYGBw
Two MSimdBox instructions with the same SimdType must have identical template
objects, but they do not need to have identical initial heaps. One could
represent a pre-tenured SIMD object allocation.
MozReview-Commit-ID: ITymizUKjlN
Add a function SimdTypeToString used for logging purposes only.
Include the exact SimdType and initial heap when logging MSimdBox and
MSimdUnbox instructions.
MozReview-Commit-ID: 1WbwaQgMLqm
Mostly just declaring globals that Cu.imports defines but there are some actual
bugs here that have been fixed as well as one test that just never ran because
of a hidden exception.
MozReview-Commit-ID: J6uIpYp8ANx
This defines a few additional globals but also turns on the browser environment
for everything in browser and toolkit. This may lead to some false negatives
but we have lots of code that runs in a browser context so in the name of
getting rules turned on I think this is a useful step.
MozReview-Commit-ID: BdWouZGK6d
This adds more of the scripts that browser.js relies on and also makes
browser-chrome head files import the browser.js globals.
The MOZ_JSDOWNLOADS block in contentAreaUtils only seems to hide a single
function, I don't see any need to keep hiding that now we're on by default.
MozReview-Commit-ID: 5zvF3JtJrZG
To properly lint XBL files we need to support things like import-globals-from
and other ESlint comment directives so we have to pass comments through to the
code blocks that ESlint parses. Unfortunately the way the XBL processor works
now is by passing a separate code block for every method/property/etc. in the
XBL and ESlint doesn't retain state across the blocks so we would have to prefix
every block with every comment. Instead this change makes us output just a
single block that roughly looks like this:
<comments>
var bindings = {
"<binding-id>": {
<binding-part-name>: function() { ... }
}
}
This has some interesting bonuses. Defining the same ID twice will cause a lint
failure. Same for the same field in a binding. The line mapping is a little
harder and there are still a few lines that won't map directly back to the
original file but they should be rare cases. The only downside is that since
some bindings have the same binding declared differently for different platforms
we have to exclude those from linting for now.
MozReview-Commit-ID: CAsPt5dtf6T
While working on turning on no-undef I discovered that the various rules we
have for defining globals are a little inconsistent in whether the files they
load recurse through import-globals-from directives and none of them imported
eslint globals directives.
I think we're better off putting all this global parsing code in a single place
rather than spread across multiple rules. Have one rule to turn it on for
parsed files and one function to load globals from other files and make them
share most of the code so we won't get inconsistent. If we find us needing to
turn on/off individual features we can figure out a way to do that in the
future.
This patch does that, the globals.js file does all global parsing with a shared
object that receives events from the AST, either through from an ESlint rule
or from a simple AST walker using estraverse.
MozReview-Commit-ID: 9KQZwsNNOUl
