search

desihub / imaginglss

Code for doing large-scale structure studies with the imaging data.
Other
6 stars 2 forks source link

Distribution and Installation #7

Open rainwoodman opened 8 years ago

rainwoodman commented 8 years ago

Support pip install.

Do we go through the pypi?

rainwoodman commented 8 years ago

@martinjameswhite @ekitanidis

What is your thoughts on this?

uploading to pypi will allow people to install / update with pip install --user --upgrade imaginglss

This will force us to follow the 'develop/'release'/'distribute' cycle.

Either Ellie or me can open an account on pypi and register imaginglss there.

weaverba137 commented 8 years ago

As a step toward this, I recommend that you modify the package infrastructure to follow, e.g., desispec.

In general we are trying to make DESI packages pip-installable, but that doesn't mean you have to publish them on pypi (though we could do that in the future).

If you follow the template of the packages desispec, desiutil, or desitemplate, you will automatically make your package pip-installable. The primary difference among these packages are what dependencies are needed to satisfy Travis tests.

rainwoodman commented 8 years ago

Thanks @weaverba137 . After discussing with @martinjameswhite, we decided against push imaginglss towards a formalized 'release' cycle at this early stage.

The current model (pip install from the source + directly launching from the source) looks sufficiently flexible and we will look into desi-template when the current model becomes a limiting factor.