| Commit message (Collapse) | Author | Age | Files | Lines |
| | |
|
| | |
|
| | |
|
| |
|
|
| |
This will allow us to upgrade to the latest version of rusqlite.
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
Adding an exception for now, as per #1608
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixes the `minimal-versions` failure:
```
error: package ID specification `[email protected]` did not match any packages
Did you mean one of these?
[email protected]
```
(we don't actually need to un-downgrade `futures-lite` anymore)
But we *do* need to un-downgrade `event-listener`
(`async-global-executor` needs `event_listener::EventListener` to be `UnwindSafe`,
and in `event-listener 3.0.0`, they are not)
|
| | |
|
| | |
|
| |\
| |
| |
| |
| |
| |
| | |
Fix reproducible build CI job for macOS
Closes #1394 and #1507
See merge request tpo/core/arti!2377
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
osxcross tries in various ways to detect which version of the compiler
we have, which version of the OS, which C++ standard library we use, and
many other things. With this variable set, osxcross seems able to detect
the environment provided by the SDK that we use.
The error messages from osxcross is as follows:
osxcross: error: cannot find libc++ headers
osxcross: error: while detecting target
This origins from its internal target detection code in target.cpp of
osxcross. This file contains a signficant amount of the detection for
paths of different versions of the macOS SDK's and macOS versions. All
of this is truly cursed.
In osxcross's main.cpp, the variable is read in:
if (char *p = getenv("MACOSX_DEPLOYMENT_TARGET")) {
target.OSNum = parseOSVersion(p);
unsetenv("MACOSX_DEPLOYMENT_TARGET");
}
and setting it thus forces the OS version to be parsed from the variable
instead of its (in our case failed) attempt at detecting the version
itself.
Fixes: tpo/core/arti#1394
|
| | | |
|
| | | |
|
| |\ \
| | |
| | |
| | |
| | | |
Run tests of every crate, with all features disabled
See merge request tpo/core/arti!2350
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
We can't do this for every crate.
I looked at what is now matrix-check to see if I wanted to use any of
the code. But it seems too entangled with its particular purpose, and
has a lot of embedded knowledge of our crates' features. I found it
sufficiently far from what I wanted that I decided on a fresh script.
|
| | |/
| |
| |
| |
| | |
There's a TODO asking for `check` to be changed to `test`. And we're
about to invent a thing that does, sort of, the `test`.
|
| |/ |
|
| |
|
|
|
| |
Now that it's gone, we no longer need to bless its MPL-2.0 usage
or excuse it from cargo-audit.
|
| |\
| |
| |
| |
| |
| |
| | |
Test cbindgen correctness in CI
Closes #1502
See merge request tpo/core/arti!2320
|
| | |
| |
| |
| | |
Previously it didn't actually fail if there was a change.
|
| |\ \
| |/
|/|
| |
| | |
Add a few more tor employes to exclude from our acknowledgments
See merge request tpo/core/arti!2306
|
| | |
| |
| |
| |
| |
| | |
(Our practice has been that after one or two release cycles,
employees get left off the list of "thanks" in the changelog,
and instead get counted as among the people _doing_ the thanking.)
|
| | |
| |
| |
| |
| | |
The module is correctly documented as "only available on crate feature
restricted-discovery" without it.
|
| | |
| |
| |
| |
| | |
The script isn't handling the `cfg_attr` on `pub mod restricted_discovery`
very well, so we need to use the `additional_required` escape hatch.
|
| |/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This makes the script work on `cfg_attr`s applied to `mod` declarations
ending in `;`.
Without this change, the script fails with
```
processing tor-hsservice
res += fn(os.path.join(dir_, file))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/builds/nickm/arti/./maint/check_doc_features", line 112, in extract_cfg_attr
end = min(subline.find(pat) for pat in ' (<' if subline.find(pat) !=-1)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: min() arg is an empty sequence
```
when run on code with `cfg_attr`s applied to `mod` declarations.
Note: this change just improves the UX a bit, but doesn't actually fix
the issue: the script still isn't able to handle `cfg_attr`'d `mod`s: it
assumes all `#[cfg_attr(docsrs, ..)]` statements are applied to
feature-gated `pub use`s, and if there aren't any (such as in the case
of `cfg_attr`d modules, it assumes the feature gating is missing.
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
This is a bit complicated since:
- Using cbindgen with macro expansion requires a nightly rust:
so, we have to look for one.
- There are some cbindgen warnings which I cannot find any way to
suppress, so instead of giving all warnings, it seems better to
give a diff from the old list of warnings to the new list.
|
| |
|
|
|
|
| |
This invalidates the cache. This may not be strictly necessary, but
it will make sure that the new pinned image is used in the CI run *for
this MR*.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
These are not hardcoded, whatever they are, so they're not the
principal thing we're trying to prevent. Previously, this would trip
on this in arti.git:maint/cargo-publish:
maint="$(dirname "$0")"
...
# shellcheck source=maint/crates-io-utils.sh
source "$maint/crates-io-utils.sh"
Ideally we'd teach this script to be able to check and maintain
includes for scripts that aren't in common/, but I think that's a task
for another day.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
set -o pipefail is defused by the if. And we should properly check
the expected statuses from git.
If git isn't found, this script would otherwise spuriously think
everything is fine.
|
| | |
|
| | |
|
| |
|
|
|
| |
This changes from `apt install` (which is not supposed to be used in
scripts) to `apt-get install`.
|
| |
|
|
| |
And delete our old version of the script.
|
| |
|
|
|
|
|
| |
See docs in maint/common/via-cargo-install-in-ci
This is not the same id as in our maint/via-cargo-install-in-ci,
which is deliberate. Let's flush this cache, just in case.
|
| |
|
|
|
|
|
|
| |
* Change all in-tree reference
* Delete our copy of script, which now lives
in the rust-maint-common subtree.
* Leave a symlink behind, so that old git hooks that people
(IMO possibly unwisely) installed, still work.
|
| |
|
|
| |
As recommended by the README there
|
| |\
| |
| |
| |
| |
| |
| |
| | |
'be8ec72c0b0ced653c618af564387079d9ea8e5f'
git-subtree-dir: maint/rust-maint-common
git-subtree-mainline: c05af7edf8f715b31d90d128ba6078a8094f440c
git-subtree-split: be8ec72c0b0ced653c618af564387079d9ea8e5f
|
| |
|
|
| |
We must make a symlink so things can find the scripts in maint/.
|
| |
|
|
|
| |
These are not parts of arti per se, and we don't especially want to
track their test coverage.
|
| |\
| |
| |
| |
| |
| |
| | |
Check cargo categories against crates.io
Closes #1481
See merge request tpo/core/arti!2256
|
| | | |
|
| | |
| |
| |
| | |
What a palaver.
|
| | | |
|
| | | |
|