Adjust star data collection based on repo popularity
KartikChugh opened this issue · comments
Testing on google/guava (38304 total stars) crashes and yields a "Please try again in a few minutes. This might not work for very famous repository" message.
I would recommend not collecting star count on a granular basis for very popular repositories. Instead, consider collecting weekly data. In the case of guava, which has existed for ~6 years, you'd have around 300 data points.
it collects only when asked... we don't keep refreshing things no one asks for, basically.
the project currently runs on the free tiers of redis and heroku, so not much I can do there for now.
that said, closing as wontfix.
I'm not sure I get your answer; I think I worded my suggestion incorrectly. I'm saying collect less datapoints when asked for repositories with a huge history. That way the service won't crash