Age | Commit message (Collapse) | Author |
|
* update version and changelog for release
* fix a brittle test that depended on rules python internals. It fails depending on which Bazel you use, even though the rule is doing the right thing.
Force merge so I can get the release to unblock last minute bazelcon demos.
|
|
Add example of new ctx.var and $(var) usage in file names.
Update common docs
- fix since regexp
- use 0.8.0 for since
- remove unneeded load
|
|
* Make sh tests portable
|
|
* Create basic bzlmod setup for rules_pkg.
- Shows it working for one example
- Has only runtime deps
- rpm and git toolchains not done yet
- Still not sure how to get the external repo test working
- Make platforms and stardoc dependency deps.
|
|
|
|
* Fix config_setting visibility failure on bazel CI
See https://github.com/bazelbuild/bazel/issues/12933.
Repro: `$ USE_BAZEL_VERSION=a05276fea75d47370b363125a074c38cb2badc74 bazelisk build --nobuild --incompatible_config_setting_private_default_visibility //src/main/java/...`
Discovered in failing Bazel CI with `--incompatible_config_setting_private_default_visibility` flipped
|
|
When we merge generated docs into the final form, convert
@since(text) to emphasized (currently italic) text.
This is not intended to be perfect. It is just to get the concept out
there to start playing with it. Ideally, StarDoc will eventually support
@since natively and we can delete this.
|
|
* Allow $(var) substitution in filenames and include everything in ctx.var in the substitution dictionary.
Fixes #20
|
|
* Adjust tar test to show #297
I tried to revert many pieces of code which should have fixed the problem, but could not reproduce 297. I'm fairly confident the root cause was eliminated a while ago.
|
|
* Cosmetic. Improve the error messageing on duplicate files in check_dest.
* lintify
|
|
* First cut at runfiles in pkg_* rules
* allow long paths in write_manifest to aid debugging
|
|
|
|
* Create an example of the kind of link tree Node users muse create
|
|
|
|
No new feature, I just want the test to make sure we don't backslide on this.
|
|
* fix #612
* check that directories get the right mode
|
|
|
|
* Properly format the deb Description field, fix format of changes file.
A combination of fixes and then tests for the behavior:
- Stop text wrapping the description. The "displayer" should do the wrapping.
- Create the changes "Description" field in the correct format.
- Do not allow newline in single line fields.
- Add leading space to continuation lines in multiline fields.
https://www.debian.org/doc/debian-policy/ch-controlfields.html
Fixes: #522
* do not test description on windows
|
|
This means that:
1. you'll get the same format no matter what version of Python you have.
At Python 3.8 the default changed from GNU to PAX.
2. The default will be suitable for building Debian packages containing
long file names.
A followup PR may add the capability to allow PAX tar writing, but I
do not know the urgency of that at this time, so that is a future
feature request.
Fixes #216
|
|
Indicates that WORKSPACE setup can be found in the release notes.
|
|
|
|
|
|
|
|
Add 'isexe' check to zip tests.
|
|
|
|
* Fix docstring to shut up buildifier
|
|
isexe checks that the file is r?xr?xr?x. That is, it just
tests for read and execute, without looking at the write bits.
The need is subtle. Some builds are local, and the executables
often come back as 0o755, while others are remote, returning
an immutable object with mode 0o555.
|
|
|
|
|
|
another revert
|
|
another CSS change
|
|
* try to skin the docs
* more css
|
|
|
|
|
|
* Support for setting executable bits.
- mostly fix #96 by finding executables and setting mode==755 on them
This new feature can not detect all executables. Bazel does not have the right
capability to make that easy. This seems to get most binaries, except shell.
You will have to wrap those in pkg_files to set the mode bits.
|
|
This change adds logic and tests for substituting values from
`pkg_zip#package_variables` into `pkg_zip#package_dir`. This behavior
largely matches `pkg_tar` (`pkg_zip` does not have an equivalent to
`pkg_tar#package_dir_file`, but they are otherwise the same).
|
|
* update README and docs for 0.7.0
|
|
|
|
* Make the .bzl files available as input to a bzl_library, without having to resort to the invasivness of a shadow tree of just the .bzl files without the other distributable files.
|
|
https://github.com/bazelbuild/rules_pkg/issues/572 - Debian policy only allows the Description field to be multiline.
|
|
* Refactor zip tests for more granularity.
The single zip_test was getting too big. I want to start creating
tests for more cases, and want to reuse the comparator tooling.
* order methods the right way
|
|
We still never create a lone '.' as a directory in the archive. That is
intentional
There are a few parts to this:
- Remove the root_directory logic from tar_writer. It was a needless feature
inherited from an ancient implementation. It was not used by anything except
the tests.
- Stop re-normalizing paths in tar_writer. It should do what it is asked. The
implication is that all the test which accounted for test data which which
contained tar files with paths like './a', will now keep the './'.
- Add a test to show prefix_dir == './' does work.
RELNOTES: pkg_tar no longer prefixes paths with './'. You can use `package_dir` with an explicit "./" to get the old behavior.
Fixes #50
|
|
|
|
* Move artifact tree handling to process_label_list.
- Use it from pkg_zip
- Move processing of files from DefaultInfo to process src
- eventually we could move tree handling from process_label_list to
process_src
- Add more docstrings.
This is a precursor to adding executable and runfiles handling to all rules.
|
|
* split apart archive.py to archive and tar_writer
No behavior change. Just make the division cleaner.
|
|
|
|
|
|
|
|
Remove the obsolete and mostly ignored pkg/README.md.
|
|
|