- Update to newer version of stardoc.
- Make lib/selects.bzl stardoc compatible. There must be a block description before Args:
Docs now generate with
bazel build --incompatible_remap_main_repo docs:*
See https://github.com/bazelbuild/bazel/issues/7130 for reasons.
* Make sets.bzl point to new_sets.bzl instead of old_sets.bzl
new_sets.bzl and old_sets.bzl should be removed in the following skylib release.
Fixes #155.
* update and rename test suites
This rule is an alternative for genrule(): it can
run a binary with the desired arguments,
environment, inputs, and outputs, as a single
build action, without shelling out to Bash.
Fixes https://github.com/bazelbuild/bazel-skylib/issues/149
native_binary() wraps a pre-built binary or script
in a *_binary rule interface. Rules like genrule
can tool-depend on it, and it can be executed with
"bazel run". This rule can also augment the binary
with runfiles.
native_test() is similar, but creates a testable
rule instead of a binary rule.
Fixes https://github.com/bazelbuild/bazel-skylib/issues/148
RELNOTES[NEW]: The new `native_binary()` and `native_test()` rules let you wrap a pre-built binary in a binary and test rule respectively.
The user can specify which line endings they want
write_file to use. This helps avoiding line ending
mismatches with diff_test.
Example: diff_test verifies that a rule generates
correct output by comparing it to a checked-in
"golden" file. Both files are text files, and the
user builds on Windows but the golden file was
written on Linux and git checkout preserved
original line endings.
Without explicitly specifying which line endings
to use, this diff_test would fail on an otherwise
good output.
With explicit line endings we don't need to check
in the golden file to git, we can just generate it
with "auto" line endings.
* Fix a number of misc issues to allow google usage of bazel-skylib
1. Missing copyright header
2. Shell test fixes to use TEST_TMPDIR to have write access to directories
3. diff_test fix to use TEST_SRCDIR
* added a comment as to why diff_test_tests is local
* ran buildifier
This new test rule compares two files and passes
if the files match.
On Linux/macOS/non-Windows, the test compares
files using 'diff'.
On Windows, the test compares files using
'fc.exe'. This utility is available on all Windows
versions I tried (Windows 2008 Server, Windows
2016 Datacenter Core).
See https://github.com/bazelbuild/bazel/issues/5508
See https://github.com/bazelbuild/bazel/issues/4319
Fixes some dictionaries to put things in the more common orders. Buildifer
used to default to doing this check and reformatting which is why the
//conditions:default got moved up in these in the first place.
All tests work with
`--incompatible_windows_native_test_wrapper`
except for the ones already broken on Windows
(//tests:analysis_test_e2e_test and
//tests:unittest_e2e_test).
See https://github.com/bazelbuild/bazel/issues/6622
Move maprule() to a private directory, to
discourage use of it. I (@laszlocsomor) am
planning breaking changes to it.
Also move private files (rule implementations) to
a subdirectory "rules/private/", to clean up the
"rules/" directory.
This PR adds two new rules: write_file and
write_xfile.
Both rules solve a common problem: to write a text
file with user-defined contents.
The problem is routinely solved using a genrule.
That however requires Bash, since genrules execute
Bash commands. Requiring Bash is a problem on
Windows.
The new rules do not require any shell.
The only difference between the rules is that
write_xfile creates an executable file while
write_file doesn't.
See https://github.com/bazelbuild/bazel/issues/4319
This PR adds two new rules: copy_file and
copy_xfile.
Both rules solve a common problem: to copy one
file to another location. The problem is routinely
solved using a genrule. That however requires
Bash, since genrules execute Bash commands.
Requiring Bash is a problem on Windows.
The new rules do not require Bash on Windows (only
on other platforms).
The only difference between the rules is that
copy_xfile creates an executable file while
copy_file doesn't.
See https://github.com/bazelbuild/bazel/issues/4319
In this PR:
- In the _resolve_locations function: use the
Bash-less ctx.resolve_tools function to resolve
the runfiles manifests and inputs of tools,
instead of using ctx.resolve_command for
the same purpose.
- In the _custom_envmap function: no longer
resolve $(location) references when
creating the envvars from custom_env, because
those references were already resolved in
_resolve_locations.
The ctx.resolve_tools() method was added in this
PR: https://github.com/bazelbuild/bazel/pull/7139
See design doc there.
- Retire the use of travis for testing.
- Update buildkite config:
- Enable buildifier testing (close to what was on travis, and is being improved.)
- Update the config to now test the latests release and the latest green, replacing
the "head" testing on travis.
* buildifier 0.22.0 includes files names in diffs.
* Simplify buildifier runs even more.
Since exit status are more consistently set and output is generally better,
don't bother being as fancy and just print a simple line as lead in incase
issues are found, and use the exit status to print directions on what to
do if something comes up.
Targets of this rule verify that targets can be analyzed successfully.
This is similar to build_test, except no actual action execution of
the underlying targets occur. analysis_test essentially verifies that
`bazel build [targets] --nobuild` passes.
* Add rules to the test_deps target.
Likely needed for anyone doing integration tests using skylib.
* Default public visibility and only tag things that are private.
