finalizing paths.json spec

[soapy1]

Creating a new pr to address the code review from #1501 (review)

This is dependent on conda/conda#3887 being merged

[kalefranz]

1. Remove .name from the line above.
2. The class below is from conda._vendor.auxlib.entity. Let's put it in conda_interface.py.

class EntityEncoder(JSONEncoder):
    # json.dumps(obj, cls=SetEncoder)
    def default(self, obj):
       if hasattr(obj, 'dump'):
          return obj.dump()
       elif hasattr(obj, '__json__'):
          return obj.__json__()
       elif hasattr(obj, 'to_json'):
          return obj.to_json()
       elif hasattr(obj, 'as_json'):
          return obj.as_json()
       return JSONEncoder.default(self, obj)

3. Add a __json__ method to NodeType that returns self.name. See other enums in conda/base/constants.py for examples.

[kalefranz]

Above there's a line

if parse_version(conda.__version__) >= parse_version("4.2"):

We should also have a section for 4.3 where we actually do the real imports from conda/exports.py. Then add this new stuff under the else after 4.3. That way it's easy to rip out in the future.

[kalefranz]

LGTM

[soapy1]

I think these errors are the same kind that we were seeing in conda and has to do with updates to a new version of requests

[msarahan]

Please merge master to resolve the conflict. Let me know if you have any questions.

[kalefranz]

@soapy1 let's get master merged in, see if we need fixes, and then get it merged

[kalefranz]

Wait, how did soapy1#14 get merged, but I don't see the commit?

[soapy1]

Oh, it looks like the changes aren't in here either. I had to rebase on master to resolve some merge conflicts, could that have caused it?
I'm going to merge it in again.

[msarahan]

Appveyor failure: https://ci.appveyor.com/project/ContinuumAnalyticsFOSS/conda-build/build/1.0.1069/job/0v2qc287b1gnl2d5#L1063

Looks like python 2.7 does not define os.link in windows.

[soapy1]

conda canary test is failing because conda 4.3 isn't actually exporting FileMode this must have slipped through the cracks.
Submitted a pr to fix this: conda/conda#4080

[kalefranz]

Looks like python 2.7 does not define os.link in windows.

I guess it's not yet so, but here it seems appropriate to export create_link() from conda/gateways/disk/create.py.

[soapy1]

@kalefranz I tried to get around that problem by having another test on windows that doesn't need link: https://github.com/conda/conda-build/pull/1535/files#diff-ba96ca525f7bacd1a0ee8b10d71d6b2bR63
Do you think this is a sufficient test, or should I import create_link from conda?

[msarahan]

Might want a platform skip on things other than windows. Just invert the logic from your test one up.

[msarahan]

Current windows test failures are other things that I'm looking into. This seems fine and ready to go to me.

[msarahan]

Thanks @soapy1 - merging.

[github-actions]

Hi there, thank you for your contribution!

This pull request has been automatically locked because it has not had recent activity after being closed.

Please open a new issue or pull request if needed.

Thanks!