Crate failed to build

[onur]

Docs.rs may have failed to build some crates for various reasons (missing dependency on build machine or bad configuration etc).

You can report them into this issue.

[azerupi]

https://docs.rs/crate/nickel/0.8.1

[onur]

There is a problem on crates that have url-0.5.9 dependency. If you check the build log cargo is returning error with: url-0.5.9/Cargo.toml:14:3 expected a newline after table definition. This is a warning when we use standalone cargo, but it is returning an Err as a library. I wasn't able to figure out how to bypass this.

[azerupi]

Indeed, I have seen this warning in my own crates too before. I am not sure there is a way around it, except maybe using a fork of url-0.5.9 with the fix and using Cargo replace. I wouldn't consider that a solution though 😉

I have also seen crates depending on compiler internals that did not compile with the particular nightly version used. diesel_codegen for example. But I guess making this work is a non-goal.

[dimbleby]

https://docs.rs/crate/c-ares-sys/ (and as a result, https://docs.rs/crate/c-ares/ too).

Need to install libtool to compile the C library being wrapped.

Or, install c-ares directly so that we find the package and don't need to compile our own version - either is supposed to work.

[cuviper]

https://docs.rs/crate/probe/ appears to have only attempted a few older versions, which don't work with current nightly, but the current 0.1.6 should be fine.

[onur]

@cuviper thanks for reporting. Looks like docs.rs returned IoError when trying to build probe 0.1.4+. I'll fix it tomorrow morning.

[diwic]

First - cool stuff! :-)

Second, about https://docs.rs/crate/dbus/0.3.4

Package dbus-1 was not found in the pkg-config search path.
Perhaps you should add the directory containing 'dbus-1.pc'
to the PKG_CONFIG_PATH environment variable
No package 'dbus-1' found

https://docs.rs/crate/alsa/0.1.2 fails with the same error, but missing alsa.pc instead of dbus-1.pc.

Is there some way for me as crate author to indicate that docs.rs needs to install these before compiling my crates? Because this seems like it's not going to scale very well as the number of crates grow with the fast-growing Rust ecosystem...

[lucab]

Thanks for running this service!

I noticed that systemd crates fails to build because of missing C development files: https://docs.rs/crate/systemd/0.0.10

This requires libsystemd-dev (or equivalent) on the building machine.

By the way, what is the OS of the building environment here? Does it have any package manager that could be re-used to install system dependencies on demand? I'm thinking about using the metadata table to encode this information to drive an external package manager.

[onur]

@cuviper for some reason cargo is not generating any documentation when we use target option with your crate. I fixed the IoError in docs.rs, probe is available now, but only default platform worked.

@diwic dbus and alsa building fine now.

@lucab docs.rs is using Debian. I am planning to add docs.rs metadata table to Cargo.toml, so people can define extra features and system dependencies.

BTW I installed systemd development files but looks like systemd crate (or one of its dependency) doesn't work with recent nightly.

[nagisa]

@onur I see you’ve somehow ended up running build.rs stuff when generating documentation. That’s generally unnecessary for majority of crates (notable exceptions are the stuff which use syntex/lalrpop-like scheme to generate code) and plain rustdoc should generate documentation for these crates just fine without any of the dependency libraries existing.

[onur]

@nagisa I am not sure how to achieve that but I am using Cargo::ops::compile_pkg to generate documentation. This function is building docs.

[nagisa]

@onur I filled a rust-lang/cargo#3053.

[hoodie]

https://docs.rs/crate/notify-rust/3.1.1 failed to build, due to the same problems as dbus is there a way to manually trigger rebuilding?

[onur]

@hoodie docs.rs has rebuilt every version of notify-rust.

I was planning to add a button to automate rebuild requests but I changed my mind. People can abuse this kind of feature.

[hoodie]

Perhaps there is a way to authenticate as the crate owner via github?

[msierks]

https://docs.rs/crate/lua/0.0.9/

[onur]

@msierks lua crate is looking for lua 5.3.3, and Debian only have lua 5.3.1. There is not much I can do in this case.

[hoodie]

this raises the question, do you build each crate completely or just run cargo doc? And if the latter, does cargo doc really need to have all build dependencies? This looks like a potential upstream improvement. Does it @steveklabnik?

[onur]

@hoodie cargo doc is building crate first.

[msierks]

@onur hmm, well it appears the next release rust-lua will have lua 5.3.3 bundled with it. Someone correct me if I'm wrong.

[steveklabnik]

And if the latter, does cargo doc really need to have all build dependencies?

Yes, because in order to generate docs, it has to build the library, thanks to things like cfg.

[bluss]

Thanks for making a great service. https://docs.rs/crate/nodrop-union/0.1.8 simply needs a newer nightly to build. Will it be retried automatically?

[onur]

@bluss currently there is no way to trigger rebuilds automatically. I'll upgrade rustc to latest nightly tomorrow.

[dimbleby]

Any update on this?

[onur]

@bluss nodrop-union have docs now.

@dimbleby sorry I didn't see your first message, c-ares-sys and c-ares should work fine now.

[dimbleby]

Great, many thanks!

[Robbepop]

docs.rs didn't recognize that I updated my crate Prophet from 0.0.1 to 0.0.2. It is an important update since many things (also documentation) have changed and because I yanked the old version.
docs.rs didn't list my crate in its recent released list, so I guess it was dropped or simply missed.
Is there anything I can do? At the moment I heavily depend on the great documentation generation of docs.rs. :)

[onur]

sorry @Robbepop, I think I accidentally removed prophet from build queue while I was trying to remove google-* crates because of #46 yesterday. It is fixed now.

[KenSuenobu]

Looks like the crate for glfw is failing to build. Did you guys not install cmake?

[mcobzarenco]

Also blingfire is failing because cmake is missing.

[hlb8122]

https://docs.rs/crate/bitcoin-zmq/0.1.0/builds
fails to compile due to libzmq3-dev missing?

[2ndTaleStudio]

hey there,
my crate: https://docs.rs/crate/ruspiro-boot/0.1.1 failed to build as I've set the default target to be "armv7-unknown-linux-gnueabihf", and the log fail complains about not installed compiler?:
stderr: "\n\nerror occurred: Failed to find tool. Is `arm-linux-gnueabihf-gcc` installed?

The Cargo.toml part to control doc.rs behaviour:

[package.metadata.docs.rs]
all-features = true
default-target = "armv7-unknown-linux-gnueabihf"
rustc-args = [
    "-C target-cpu=cortex-a53",
    "-C target-feature=+a53,+fp-armv8,+v8,+vfp3,+d16,+thumb2,+neon"
    ]

[dtolnay]

This build failed due to hitting rust-lang/rust#61804:
https://docs.rs/crate/syn-next/1.0.0-rc3/builds/178036

It looks like the build was with rustc 1.38.0-nightly (95b1fe560 2019-07-20). Would it be possible to update the nightly to pull in the fix for that bug? It was fixed in rust-lang/rust#62086 in nightlies newer than 2019-07-28.

[najamelan]

rustc 1.38.0-nightly (95b1fe560 2019-07-20) also fails to build futures-preview 0.3.0-alpha.18 and crates that depend on it, for issues unrelated with the above mentioned ICE, so an update of the nightly would be great!

[najamelan]

I see the rustc version has been updated, but it looks like crates that have failed before were not rebuilt after the update. (https://docs.rs/crate/futures-io-preview/0.3.0-alpha.18)

[cecton]

error[E0514]: found crate `bitvec` compiled by an incompatible version of rustc
  --> /home/cratesfyi/.cargo/registry/src/github.com-1ecc6299db9ec823/mbrman-0.1.0/src/lib.rs:72:5
   |
72 | use bitvec::prelude::*;
   |     ^^^^^^
   |
   = help: please recompile that crate using this compiler (rustc 1.39.0-nightly (bea0372a1 2019-08-20))
   = note: the following crate versions were found:
           crate `bitvec` compiled by rustc 1.36.0 (a53f9df32 2019-07-03): /home/cratesfyi/cratesfyi/debug/deps/libbitvec-037542c063cb88fa.rmeta

error: aborting due to previous error

(logs)

I got this error but I cannot reproduce locally using rustc 1.39.0-nightly (4295eea90 2019-08-30)

[nbdd0121]

• https://docs.rs/crate/mnl-sys/0.1.0
• https://docs.rs/crate/mnl/0.1.0
• https://docs.rs/crate/nfq/0.1.0

Failed due to missing libmnl-dev

[driftluo]

https://docs.rs/crate/tentacle/0.2.4
https://docs.rs/crate/tentacle-secio/0.1.10
https://docs.rs/crate/tentacle-ping/0.3.6
https://docs.rs/crate/tentacle-identify/0.2.7
https://docs.rs/crate/tentacle-discovery/0.2.6

These libraries have encountered a problem. They all have 2 features to choose from, but they can't choose all or not noe. The default selection is empty. I don't know how this is expressed in toml metadata, so the doc generation for these libraries has failed.

#[cfg(all(feature = "flatc", feature = "molc"))]
compile_error!("features `flatc` and `molc` are mutually exclusive");
#[cfg(all(not(feature = "flatc"), not(feature = "molc")))]
compile_error!("Please choose a serialization format via feature. Possible choices: flatc, molc");

---

I know how to config it now. https://docs.rs/about

[agerasev]

Hi, thank you for running Docs.rs!

https://docs.rs/crate/clay-core/0.1.0

The build failed due to error: could not find `format_args` in `core` while building ocl. The cause might be similar to rust-lang/rust#63687.

Maybe the problem is in rustc 1.39.0-nightly (618768492 2019-09-05) used by Docs.rs. I cannot reproduce this error locally using rustc 1.39.0-nightly (f0b58fcf0 2019-09-11).

[pietroalbini]

@nthend updated the nightly on docs.rs and manually queued a build for your crate!

[Riateche]

Hi! It'd be nice to have docs for my Qt crates (qt_core, qt_gui, qt_widgets, qt_ui_tools, qt_3d_core, qt_3d_render, qt_3d_input, qt_3d_logic, qt_3d_extras, qt_charts, qt_qml) on docs.rs. Can you install the corresponding Qt development packages in your build environment? Qt >= 5.9.7 will do.

Ideally, the RITUAL_RUSTDOC=1 environment variable should also be set. This will enable documentation of all available items, independent on the current Qt version. This is basically a hack to achieve the behavior of #[cfg(rustdoc)] that is not available on stable, unfortunately. Or, if you don't want to pass ritual-specific variables, you may set an env var like IS_DOCS_RS=1, and I'll read it in my build script.

Build log

[Hexilee]

As this docs building.

It's obvious that docs.rs does not execute proc-macro-attribute, because each async fn in trait will be transferred into a normal fn by proc-macro-attribute.

Cloning this repo and executing cargo doc --open --all-features locally works well.

[rodrimati1992]

abi_stable and abi_stable_derive are not in the queue/Recent failures/recent releases lists,even though updates of other crates released afterwards were built.

Update
While the 0.7.0 version wan't built,the 0.7.1 built just fine.

[manuthambi]

I created a new crate big_enum_set. But the docs are not showing up on docs.rs. cargo doc works locally.

Perhaps related?

[jeremystucki]

We have the same issue with mockiato. No build has been started for version 0.9.4.

[Cerber-Ursi]

Fresh build of proc-macro2 has failed too. From the log, it looks like some misconfiguration in the docker create command - the argument RUSTFLAGS=--cfgprocmacro2_semver_exempt seems to be incorrect. It comes from the following section of Cargo.toml:

[package.metadata.docs.rs]
rustdoc-args = ["--cfg", "procmacro2_semver_exempt"]

I wasn't able to find another example yet, but possibly this will crash docs building for any crate with similar configuration.

[pietroalbini]

@Cerberuser yep, that's indeed a bug in docs.rs. Opened a PR with a fix: #427. Once the PR is merged we will rebuild that crate.

[dignifiedquire]

pgp is failing: https://docs.rs/crate/pgp/0.2.2/builds/187802, but cargo doc runs fine locally

[cecton]

How do I know actually if it's a doc-rs problem or not? A single issue on GitHub seems pretty inconvenient for seeking some help...

[pietroalbini]

@dignifiedquire rebuilt pgp 0.2.2! It was a docs.rs error, but it was fixed a few days ago.

@Cerberuser rebuilt proc-macro2 as well after the fix for the issue landed, but there was a compilation error: https://docs.rs/crate/proc-macro2/1.0.5/builds/190515.

[jeremystucki]

@pietroalbini Could you trigger a build for mockiato = "0.9.4" as well?

[Cerber-Ursi]

@pietroalbini Couldn't reproduce the failure myself, so pinging the author - cc @alexcrichton

[pietroalbini]

@jeremystucki queued both the failed mockiato versions (0.9.3 and 0.9.4) again!

[pietroalbini]

Hi everyone!

It became apparent to us that this issue doesn't really scale: people posting here start receiving useless notifications and need to manually unsubscribe from the thread, and it's hard for us to keep track of which crates are fixed and which need action from our end.

We now recommend everyone to create a separate issue for each crate, and to aid that we added issue templates for broken crates and sandbox limit increase requests, changing the website to point to them. This issue will be also locked to prevent further reports here.

If your crate still fails please open a new issue and we'll look into it!