Conda dependencies that require cchardet<2.0 out of sync w/ setup.py

[Zane Selvans]

Despite the fact that the datapackage-py setup.py script in the GitHub repository requires only cchardet>=1.0, the distributed package in the conda-forge channel for v1.6.0 still impose a maximum cchardet version of <2.0. (see conda output below). This means that datapackage-py can't be used in a software environment alongside Python 3.7.

(pudl) %0 ~/code/catalyst/pudl% conda search datapackage=1.6.0 --info
Loading channels: done
datapackage 1.6.0 py_0
----------------------
file name   : datapackage-1.6.0-py_0.tar.bz2
name        : datapackage
version     : 1.6.0
build       : py_0
build number: 0
size        : 40 KB
license     : MIT
subdir      : noarch
url         : https://conda.anaconda.org/conda-forge/noarch/datapackage-1.6.0-py_0.tar.bz2
md5         : 301bccf1d84bac7e45aec0d500f400cc
timestamp   : 2019-04-11 14:33:33 UTC
dependencies: 
  - cchardet >=1.0,<2.0
  - click >=6.7,<7.0
  - jsonpointer >=1.10,<2.0
  - jsonschema >=2.5,<3.0
  - python
  - requests >=2.8,<3.0
  - six >=1.10,<2.0
  - tableschema >=1.1.0,<2.0
  - tabulator >=1.3,<2.0
  - unicodecsv >=0.14,<2.0

[Zane Selvans]

This definitely seems weird, given that the setup.py version tagged for 1.6.0 definitely doesn't have these requirements.

[roll]

@zaneselvans
It's really weird because datapackage has had this <2 requirements lately - 411675d (it was <3 so not sure from where this <2 is coming)

Is it possible that some underlying dependencies is not updated?

[roll]

I see it's coming from tabulator@1.3 which is a dependency of datapackage. I'm releasing datapackage@1.6.2 with updated tabulator dependency.

It's on PyPi now but not on Anaconda yet.

[Zane Selvans]

Yay! Great. Glad it's figured out.