Back-end junior uawebchallenge work.
- Ruby as a language, Sinatra as a simple DSL for website tasks.
- Mongo as a database, Mongoid as an adapter.
- Rspec for tests, VSR to cache web requests during testing.
Any request result will be saved to database at first attempt. To save it to database you need to interact with this site usual way. I didn't create any scheduled works due they are useless with Heroku free account (but it is good solution for caching our data anyway): application will go asleep after thirty minutes of idle and nothing will be done. So, where you will make first request, please, wait a little bit for slow free Heroku app will done all the work. There is four different requests that can be performed each day, but only three will be different:
- Apps created last week / apps' activity for last week
- Apps created last week / apps' activity for last month (result will be the same as first one, this state is checked)
- Apps created last month / apps' activity for last week
- Apps created last month / apps' activity for last month
This task shows perfectly typical N+1 query problem to external API with huge side computing. I may be wrong but I think that such kinds of clever/tricky tasks fits perfectly when we can design our API. Anyway, the task is comlete.
I made a research and communicated with github support. Here is their answer:
'There's no equivalent in the API for the data shown on the Trending page. Also, the trending page shows only stars, not commits or contributors, as you mentioned. To answer your question -- no, there's no simple way to collect this data via the API in a small number of requests. You might need to make several requests per number per repository.'