Update Dependencies

[bradyzp]

Consider/test updating certain dependencies, particularly scipy.

The current version of scipy cannot be installed via pip on Windows, and requires some complex compilation/installation steps (better to install Anaconda or similar). Scipy v1.1.0 resolves most of these issues, but there may be some breaking changes in the API which will need to be corrected.

Other updates to consider:

• pandas 0.20.3 -> 0.23.1
• PyQt5 5.9 -> 5.11.x

Another point of consideration is whether it is desirable or serves a purpose to continue supporting CPython 3.5.x as opposed to focusing exclusively on Python 3.6+
There are a few would-be-nice features added in 3.6 such as inline variable type annotations, fstrings, etc.

[cbertinato]

Let’s try to upgrade scipy and see what breaks. I don’t think we use it so heavily that a breaking change couldn’t be fixed with a small change on our part. We can go to pandas 0.23.1. I don’t think there are any significant breaking changes for us there as well. I don’t know much about the changes in PyQt, so I will defer to you on that one. I’m also ok with going to Python 3.6. I’ve been using it with pandas since 0.19 and have not had issues. In general, I think that most intentional breaking changes were made for a good reason, and we are still small, so we can deal with them to make the upgrades fit. On Jul 4, 2018, at 04:04, Zac Brady <[email protected]> wrote: Consider/test updating certain dependencies, particularly scipy. The current version of scipy cannot be installed via pip on Windows, and requires some complex compilation/installation steps (better to install Anaconda or similar). Scipy v1.1.0 resolves most of these issues, but there may be some breaking changes in the API which will need to be corrected. Other updates to consider: - pandas 0.20.3 -> 0.23.1 - PyQt5 5.9 -> 5.11.x Another point of consideration is whether it is desirable or serves a purpose to continue supporting CPython 3.5.x as opposed to focusing exclusively on Python 3.6+ There are a few would-be-nice features added in 3.6 such as inline variable type annotations, fstrings, etc. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#68>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ATz35p5_nCAw9Zfx3xc97VV3AhdRHMVXks5uC8BKgaJpZM4VBenH> .

[bradyzp]

Hey Chris sounds good.

I dropped 3.5 from the CI configurations in my last push, and updated scipy and some other dependencies without issue.
There is some change in pandas 0.23.x that breaks two tests in the trajectory_ingestor library, so I'll keep it pinned at the current version we're using for now.

Also on that note I've added in a CI configuration to test on AppVeyor, and started running coverage reports which are published to coveralls.io

[cbertinato]

Excellent! Thanks Zach. I’ll take a look at the pandas updates when I get back next week. On Jul 6, 2018, at 04:22, Zac Brady <[email protected]> wrote: Hey Chris sounds good. I dropped 3.5 from the CI configurations in my last push, and updated scipy and some other dependencies without issue. There is some change in pandas 0.23.x that breaks two tests in the trajectory_ingestor library, so I'll keep it pinned at the current version we're using for now. Also on that note I've added in a CI configuration to test on AppVeyor, and started running coverage reports which are published to coveralls.io — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#68 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ATz35paHdKE5dGe_Co1p2aVNldfI1GClks5uDmdfgaJpZM4VBenH> .