Data one cannot cache - Githubissues

ropensci-books / http-testing

HTTP testing for R

https://books.ropensci.org/http-testing/

Creative Commons Zero v1.0 Universal

53 stars 17 forks source link

Data one cannot cache #23

Closed maelle closed 3 years ago

maelle commented 4 years ago

Secure data but also data one doesn't one (check the data provider is ok with data being cached for tests or could a data provider think it's resharing the data)

sckott commented 4 years ago

I'm not sure I follow your description, when you get back from vacation, please explain in more detail

maelle commented 4 years ago

Answering before I forget what I meant. 😅 I was (yes not clearly) wondering whether one could get in trouble with a commercial web service if publicly storing responses in non encrypted cassettes. Depending on the data usage guidelines of the service. E.g. Google, Twitter.

I think in the book it could mean that when mentioning encryption of cassettes, one should say it's for sensitive data (like personal data, medical data) but also for data one doesn't own (geocoding results from Google)?

sckott commented 4 years ago

thanks @maelle ! makes sense now. yes, we should add notes on this, will do

maelle commented 4 years ago

Related to #2

maelle commented 3 years ago

If you are dealing with sensitive data in your API (cf https://github.com/ropensci/qualtRics/pull/140#issuecomment-619647467) you probably don't want to save actual data.

So you'll want to either

edit the recorded files
use a sandbox service / demo service so you'd be caching demo data.

Having a sandbox is the best solution because it also means other contributors to the package get the same data.

maelle commented 3 years ago

I could show how to encrypt files with sodium, using example from gargle code. That applies to tokens but one could also encrypt cassettes although of course that is not cool for PRs from forks.