Some Travis checks are failing

[kocio-pl]

If you look inside Travis tests, like this:

https://travis-ci.org/gravitystorm/openstreetmap-carto/builds/162536783

there are multiple errors like:

Error: placenames.mss:65:4 Unrecognized rule: shield-margin

or

-:1: parser error : Document is empty

Yet the "build exited with 0".

Is it expected behavior and should ignored or there's something wrong under the hood and some action is needed?

[StyXman]

From bash's doc: The return status of a pipeline is the exit status of the last command, unless the pipefail option is enabled. If pipefail is enabled, the pipeline's return status is the value of the last (rightmost) command to exit with a non-zero status, or zero if all commands exit successfully.

The test is adding wc -l as the last command, so unless pipefail is enabled, this will always succeed. I can only wonder why is that there. Should we replace it with a redirection to /dev/null?

[matthijsmelissen]

Thanks @StyXman!

[HolgerJeromin]

@StyXman the wc is there to see if the style rules generates a combinatorial explosion.
But i dont know if someone really looks at the number and is alerted if the line count jumps :-)

[nebulon42]

Speaking of Travis tests: is there a specific reason why still 0.12.1 is used for Carto? Is that the live version? IMO it should be replaced with the latest version but targeting a specific Mapnik API version. This is now possible (since 0.16) with:

-a / --api VERSION - Specify Mapnik API version (e.g. --api 3.0.10) (default: 2.3.0)

[kocio-pl]

I think we're ready to drop MML, so 0.16.x would be necessary (live version is currently 0.16.3).

It would probably also mean more changes in .travis.yml and removing travis_check_project_files - am I right?