Closed tomschr closed 6 years ago
I'm remembering now why I implemented deploy_docs
that way. The issue is that ghp_import
is not designed to be use as a calling interface, it's a console script. Installing ghp_import doesn't actually let you import the module, the only file available from the install is ghp-import
, no file extension.
I used run simple because it used ghp_import
as it was designed to be used.
Yes, maybe it was a bad idea. ;-)
However, maybe you should add this line to your setup.py
to state a dependency to ghp_import
:
install_requires=[
'sphinx', 'ghp_import',
],
With the 2.0.0 changes, this issue is no longer applicable. Thanks for the issue none the less!
I observed you use the
ghp_import.py
script from theghp_import
repository.The interesting part is line 49 where you clone the repository with git and execute it (function
deploy_docs
). I'm not sure what the story behind this design decision is, but this is a bit, well... unusual. ;-)The problem with this approach is:
ghp_import
automatically.setup.py
and check for requirements cannot get the "full" list of requirements (becauseghp_import
is "hidden" inside your script).Is there a reason why you don't require this in the project? I would propose the following changes:
State clearly in
setup.py
that this project requiresghp_import
:If possible, don't clone the repo and run the script. Instead import it and use it as follows:
Probably the
ghp_import.py
wasn't meant to be a library but a script. Maybe the upstream project should split the script into a library part and a script part. That would make things easier for you.If the last entry from the above list doesn't work, you could just run the
ghp_import.py
script.How does it sound? :-)