Description of problem: The "macros" feature is not included in the package and is patched out of the default feature list. This causes build failures for any dependent package that relies on "macros" being enabled by default, or that explicitly enables the feature. Version-Release number of selected component (if applicable): 1.9.4-1.fc36.noarch How reproducible: Always Steps to Reproduce: 1. Build a package (in this case, rust-coreos-installer 0.12.0) that uses a serde_with macro and builds with default features. Actual results: error[E0432]: unresolved imports `serde_with::serde_as`, `serde_with::skip_seria lizing_none`, `serde_with::DeserializeFromStr`, `serde_with::SerializeDisplay` Expected results: Code compiles. Additional info: As expected, patching the dependent Cargo.toml to request `features = ["macros"]` causes a failure to resolve BuildRequires: Problem: nothing provides requested (crate(serde_with/macros) >= 1.9.4 with crate(serde_with/macros) < 2.0.0~)
Yup, the way this package is built is broken, and I would've probably noticed this sooner if it had been added to the @rust-sig group. I'm working on updating serde_with_macros and darling, but I'd rather wait for pbrobinson to respond or finally add his missing Rust packages to the rust-sig group.
FEDORA-2021-a4565346aa has been pushed to the Fedora 36 stable repository. If problem still persists, please make note of it in this bug report.
(In reply to Fabio Valentini from comment #1) > Yup, the way this package is built is broken, and I would've probably > noticed this sooner if it had been added to the @rust-sig group. I'm working > on updating serde_with_macros and darling, but I'd rather wait for > pbrobinson to respond or finally add his missing Rust packages to the > rust-sig group. Finally? You waited two days before stepping in, also at this time of the year. I don't remember why we patched it out, there was either an issue with it at the time, like a rebase of a library version that was blocked on something else or it needed massive amounts of dependencies and we didn't need the functionality. In terms of adding the package to @rust-sig.... well if you'd actually engage rather than just going any randomly breaking stuff where I get a whole lot of extra bugs and work to do like 2027326 I may actually be inclined to do so but it seems you'd just prefer to make snide comments and push stuff anyway.
Peter, I'm sorry about my choice of words. I am as frustrated with people pushing broken changes as you are, and that sometimes makes its way into the way I write comments (in addition to English not being my native language, which doesn't help things either). Regarding pushing the fix myself, I didn't think that I needed to wait for a response from your side before pushing an - at least to me - obvious and simple fix to unblock work on rust-coreos-installer. Such changes are even within the usual reponsibilities of provenpackagers, even if the package in question wasn't part of the set maintained by the Rust SIG, so I did not think about it too much. Concerning adding some of your Rust packages to the Rust SIG, I have already reminded you of that multiple times last year, and you also should have received a very nice email from me (either one of the devel-, rust-, or rust-sig- lists, or directly as Bcc recipient) about this, almost three weeks ago, where I included nice lists and even a python script to add packages to the Rust SIG without having to click through the pagure web GUI for every single one: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/message/5BTNYNCR3LGJT4QK4VO3AXOTRDQYX45P/ > well if you'd actually engage rather than just going any randomly breaking stuff where I get a whole lot of extra bugs and work to do like 2027326 I'm sorry about how that particular change was handled. It looks like eclipseo was a bit overwhelmed with work, missed a few dependent packages / things that should not be updated in some branches, and as a result pushed incomplete updates for bindgen 0.59 to stable branches: https://bodhi.fedoraproject.org/updates/FEDORA-2021-1937205ae5 https://bodhi.fedoraproject.org/updates/FEDORA-2021-b9e6038b3b As you can see, I tried my best to fix those updates after they were initially pushed by adding the missing changes and package rebuilds, but obviously I too missed some non-obvious things, the mess was already made, and a few packages got broken as a result. The experience with those too-big updates is one of the reasons why I changed the way I pushed Rust updates a bit - splitting changes into smaller chunks, and better making sure to avoid broken updates. So far, the rate of "broken updates" that were pushed has dropped to zero (and it was not high to begin with, you just seem to have been unlucky to have been hit by two issues within the last month or two).