Conversation
|
Overall, I am appreciative of these changes. Though I wonder if there is utility in keeping the removed doc tests and instead of having them wrap While we should entirely revamp that page, I don't want to lose the examples that are there. Can we either:
|
|
Oh. I don't know. Didn't think of either of those options to be honest. The removed doc-test is because these invocations won't work anymore. EIther way, I welcome changes to this PR. I'm not planning on doing more here. |
JosiahParry
left a comment
There was a problem hiding this comment.
I have one question regarding scope of the mutable assignments.
But we have a lot to add to the changelog. And should add receipts from r-devel exchanges as well.
While I agree this is a good idea for a temporary solution, I'd recommend you to re-implement the functionality using the provided APIs (or simply remove them). I think the R core team is cleaning up these "non-API"s, so it's possible there's no time until they actually remove them (I guess not all, but some), which means extendr won't compile with the |
* Update to extendr-api v0.6.0 * Go offline * Downgrade regex for building on older rust toolchain. * updates to use Extendr from main * buildignore the .vscode dir * Change the error type * Workaround bug in dev extendr * add a r-devel check * Upgrade minimum Rust toolchain required to 1.69 * Bundle extendr from extendr/extendr#809 * re-vendor for line endings * Try cleaning up header right after building. * Instead simply cleanup all headers
CGMossa
left a comment
There was a problem hiding this comment.
The CHANGELOG works for me.
|
Ubuntu CI is failing due to multithreading of the cargo tests which is poisoning the thread safety mutex. This can be fixed by #674 The windows tests are failing at extendr-api/tests/extendr_macro.rs |
|
If I rebase, then the meta data tests will succeed. |
8ba040d to
0fc50c8
Compare
I've now rebased. I don't understand what to reply to this. Is it possible to do or not? |
* Cargo.toml: update libR-sys to latest commit * added more `non-api` cfg guards * moved doctests to their own module * wp: another * wp: another one * this test is disabled, yet it still compiles.. * typo * removed more tests * update changelog * update note on CRAN * `cargo extendr fmt` [skip ci] --------- Co-authored-by: Josiah Parry <josiah.parry@gmail.com>
* Cargo.toml: update libR-sys to latest commit * added more `non-api` cfg guards * moved doctests to their own module * wp: another * wp: another one * this test is disabled, yet it still compiles.. * typo * removed more tests * update changelog * update note on CRAN * `cargo extendr fmt` [skip ci] --------- Co-authored-by: Josiah Parry <josiah.parry@gmail.com>
* Cargo.toml: update libR-sys to latest commit * added more `non-api` cfg guards * moved doctests to their own module * wp: another * wp: another one * this test is disabled, yet it still compiles.. * typo * removed more tests * update changelog * update note on CRAN * `cargo extendr fmt` [skip ci] --------- Co-authored-by: Josiah Parry <josiah.parry@gmail.com>
CRAN compliance PR.
Background: CRAN has new requirements for what is non-API. These requirements just came up and became defacto law of the land. They are not tied to any R release. We just have to deal with it.
We pulled non-API list in
extendr/libR-sys#248
This PR syncs
libR-syswith this commit.non-apifeature. It means, if you want it, you may compileextendr-apiwithfeature = ["non-api"]. This requires you to build your own bindings by way oflibR-sysas wel.non_api_tests, and in order to make that work, I had to add an entry for it inextendr-api/Cargo.toml. Note that theoptional-module in integration tests doesn't work. And it needs to also have an entry to work.required-features, but this doesn't run the test unless the feature is provided to thecargo test-call. So these tests are present, but won't run. We have to have a broader conversation about testingnon-api. For now, this is a patch, that allows us to rediscover missing features if users want them.