Add recipes for Chemfiles

[Luthaf]

A C++ library for reading/writing theoretical chemistry trajectories, with Python bindings. See http://chemfiles.github.io/ for more informations.

I added three recipes: C++ library, Python binding and meta-package. If you prefer having everything in one recipe, I can to it.

[conda-forge-linter]

Hi! This is the friendly automated conda-forge-linting service.

I wanted to let you know that I linted all conda-recipes in your PR (recipes/chemfiles, recipes/chemfiles-lib, recipes/chemfiles-python) and found some lint.

Here's what I've got...

For recipes/chemfiles:

• The top level meta keys are in an unexpected order. Expecting ['package', 'requirements', 'about', 'extra'].
• The recipe must have some tests.
• The recipe must have a build/number section.

For recipes/chemfiles-lib:

• The top level meta keys are in an unexpected order. Expecting ['package', 'source', 'build', 'requirements', 'about', 'extra'].
• The recipe must have some tests.
• The recipe must have a build/number section.

For recipes/chemfiles-python:

• Failed to even lint the recipe (might be a conda-smithy bug) 😢

[jakirkham]

Please add toolchain. This configures the compiler to use and how to use it.

[jakirkham]

Thanks for proposing these, @Luthaf. Glad to see you over here. 😄

Tried to add some addition comments in addition to what the linter included. Once we get these cleaned up a bit more, we can take a closer look.

[conda-forge-linter]

Hi! This is the friendly automated conda-forge-linting service.

I wanted to let you know that I linted all conda-recipes in your PR (recipes/chemfiles, recipes/chemfiles-lib, recipes/chemfiles-python) and found some lint.

Here's what I've got...

For recipes/chemfiles:

• The recipe must have some tests.

For recipes/chemfiles-lib:

• The recipe must have some tests.

For recipes/chemfiles-python:

• The top level meta keys are in an unexpected order. Expecting ['package', 'source', 'build', 'requirements', 'test', 'about', 'extra'].
• The recipe must have a build/number section.

[Luthaf]

I addressed most of the comments; still have to enable tests for chemfiles-lib and check for what is going on with export MACOSX_DEPLOYMENT_TARGET="" # [osx].

[jakirkham]

I know the linter is mentioning tests. There are a couple types of tests we like to see.

Running the test suite (mainly for compiled code) so make check, ctest or similar. Normally that happens in the build.

For Python libraries import tests in the test section are nice.

Some checks to make sure files (libraries, headers, configuration files, etc.) were installed where they should be.

[conda-forge-linter]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipes/chemfiles, recipes/chemfiles-lib, recipes/chemfiles-python) and found it was in an excellent condition.

[Luthaf]

I actually need to export MACOSX_DEPLOYMENT_TARGET="", because without it I get a Bus Error (segfault) in one of my tests. I will try to debug this using the binary built by conda, but I cannot reproduce with a fresh build.

[Luthaf]

The Circle-CI failure is strange: I got an error about netcdf.h file not found, even though libnetcdf is part of the tests.require entry. Any idea on the origin of this failure?

[Luthaf]

So after investigations (see the linked issue), the bus error is very weird. I am in favor of just unsetting MACOSX_DEPLOYMENT_TARGET in the build, since that make the bug disappear.

This is fixed by updating Boost, and I release the 0.6.2 version with this fix on Chemfiles.

@jakirkham, I do not get the origin of the failure on travis/circle-ci, any idea? The Appveyor issue looks strange and related to installing libnetcdf with MSVC.

[Luthaf]

Appveyor failure seems legitimate, the build log is

conda.resolve.Unsatisfiable: The following specifications were found to be in conflict:
  - cmake
  - libnetcdf 4.4.* -> feature:vc10
  - libnetcdf 4.4.* -> feature:vc14
  - libnetcdf 4.4.* -> feature:vc9
Use "conda info <package>" to see the dependencies for each package.
Command exited with code 1

I do not know what I can do here. It may be related to #1678, but I need VS14 to build the code here.

EDIT: maybe I can just disable the appveyor/windows version for now?

[Luthaf]

Ping @jakirkham? I disabled the Windows version, it can be added later.

[jakirkham]

So I'm in favor of merging.

Though I'd like to save us some time restarting CIs. Could we please put the non-meta recipes (looks like the first 2 commits in a separate PR)? Once done please ping me and I will merge ASAP.

Then we can cut this down the metapackage, which we can merge after the first two are built and deployed.

Sound good?

[Luthaf]

Sound good, I've separated the non-meta recipes in #1764

[jakirkham]

Thanks. Have gone ahead and merged that. It might be a little bit as conversion to feedstocks occurs on Travis CI ATM and they are a bit backed up as we speak.

[jakirkham]

So we ran into some issues with chemfiles-python. Have added PR ( conda-forge/chemfiles-python-feedstock#1 ) to fix them. Please take a look. Once that is sorted we should be in good shape to proceed with this.

[jakirkham]

Toggling for CI.

[Luthaf]

Should I rebase and only include the meta recipe here?

[jakirkham]

That sounds like a good idea. Sure let's do that.

[conda-forge-linter]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipes/chemfiles) and found it was in an excellent condition.

[jakirkham]

Please add skip: true # [win]. That should fix the CI failure. Also feel free to add [skip appveyor] in your commit message to avoid having to run that CI as it is not used.

[Luthaf]

Travis failure looks unrelated (502 gateway error from conda.anaconda.org), could you restart the build?

[jakirkham]

Yep. Restarted.

[jakirkham]

Thanks @Luthaf .