cljdoc

https://cljdoc.org/ & https://github.com/cljdoc/cljdoc
2018-12-21T15:21:31.000800Z

@martinklepsch Currently, how many users are using the website per minute, in average? .. just to get an idea of the scale

2018-12-21T15:22:06.001300Z

specifically, how many are using the search box per minute.

2018-12-21T15:23:14.002400Z

I am thinking that it would be easy to just not use a 100% Lucene solution for selecting the best search result, as long as the server is not under heavy load.

martinklepsch 2018-12-21T15:27:08.003500Z

I can't give you good answers to that (there's no tracking) but I don't think that there's a lot of traffic to the search currently

martinklepsch 2018-12-21T15:28:50.004600Z

@vincent.cantin what do you have in mind as an alternative to lucene? Definitely not committed to lucene but we'll need something that doesn't rely on keeping all data in memory I think

2018-12-21T15:33:01.007800Z

I am thinking that it may be doable to simply use a Clojure function for collecting the best n results, using the expressivity of Clojure instead of getting stuck on the advanced usage of Lucene, by parsing all (or a big part) of the libraries.

2018-12-21T15:35:21.009100Z

That would be a temporary solution that the end user could start to use until someone has time to use Lucene correctly.

martinklepsch 2018-12-21T18:31:59.010500Z

@vincent.cantin the issue with a Clojure function is that all the data needs to be in memory. If we're looking for a temporary solution we could also consider just using sqlite LIKE queries