search

earth-genome / coastal-valuation

A web service to assess the expected value of real estate under sea level rise
1 stars 0 forks source link

Continuous integration testing #8

Open danhammer opened 8 years ago

danhammer commented 8 years ago

Set up Travis CI for the repo. @harinisuresh is this something you'd be interested in doing? A compact way to engage. Mainly just reaching out to see if you still want to be active.

harinisuresh commented 8 years ago

@danhammer yeah I can do this! Thanks for reaching out -- still interested in being active.

danhammer commented 8 years ago

fantastic! the functions are likely to change over the next week, however, so i would hold off on writing the actual tests. Maybe just a dummy test, like 2 == 2, in the interim.

harinisuresh commented 8 years ago

@danhammer looks like to implement Travis CI in a private repo requires a paid plan (it's only free for open source repos). Shippable seems like a similar continuous integration testing service that's free for up to 150 builds per month for private repositories (https://app.shippable.com/pricing.html). Just let me know what you want to do.

danhammer commented 8 years ago

Thanks @harinisuresh. I totally forgot about the pricing restrictions. The money isn't a problem. It is, however, unclear how to properly do CI testing with Google App Engine (GAE). Can we test the web services using TravisCI or Shippable, given the weird idiosyncrasies of GAE? I asked Robin Kraft, who usually did this, and he said yes -- but it's complicated.

I think it's worth paying for TravisCI and working through these issues. I'll set this up while on the plane tomorrow. Thanks for the quick response!

danhammer commented 8 years ago

Bringing @kramachandran and @stroupaloop (who is still traveling) into this conversation. I've been doing some reading on CI testing options. The option that is lining up best with our stack seems like it is Codeship rather than Travis CI.

We have been authorized to spend the monthly fee for whatever CI testing option we choose. However, it seems like maybe we should settle on the tech stack (#5) first? We can put this issue on the back burner for the next couple of days, I think.