The way I use reddit is usually for recommendations, reviews, technical problems, etc. One main way to do this is to search a subreddit or all of reddit to see if the answer is already there. Is there a way to have a similar search on Lemmy ? I can’t seem to get this to work. It only searches for communities and not posts within them.

  • psudo@beehaw.org
    link
    fedilink
    English
    arrow-up
    10
    ·
    1 year ago

    It’s not super obvious, but if you set the communities to All in the search function it should search everything your instance is federated with

    • nobloatOP
      link
      fedilink
      English
      arrow-up
      2
      ·
      1 year ago

      Oh that’s cool. I should have specified that I was trying to do this on Jerboa. I hope that apps will implement the search functionality as well.

  • PriorProject@lemmy.world
    link
    fedilink
    English
    arrow-up
    6
    ·
    1 year ago

    I’ll echo the suggestion from another commenter to post future help questions in !lemmy@lemmy.ml. This sub is for asking lemmings about other stuff… analogous to /r/AskReddit/.

    But to summarize Lemmy’s search, it’s confusing and primitive.

    • It’s confusing because of Federation. Each instance searches its local database, which means equivalent searches from different instances are likely to return different results. Searching a popular community on a well-established instance that has been around a long while will be pretty reliable. But an instance that got started last week is going to be missing a lot of older stuff, and there’s a lot of instances like that. I recommend searching the home instance for the community rather than your home instance as a user. That will ensure that you get the full resultset for that community… though it means searching several communities may require visiting several different instances.
    • It’s primitive because it’s “just” postgres search. Most searchboxes we interact with on a day to day basis are backed by extremely sophisticated search infrastructure that has sophisticated relevance ranking that is powered by past searches… the system keeps track of how people search and modifies search results when it sees that a result is popular. Lemmy search is just using a “plain” database search, the approach to ranking is pretty simple, and it has no idea how helpful a result has been in the past. If there are few enough results that you can scroll through them all yourself that doesn’t matter much… but if a search returns hundreds or thousands of results… expect them to be ordered in a not very useful way.

    I feel like using an external search engine is probably a better approach, but that’s not super simple either. I haven’t found search strings that I’m totally happy with.

    • nobloatOP
      link
      fedilink
      English
      arrow-up
      1
      ·
      1 year ago

      Thank you! I will post any future technical questions there as suggested. Thanks for the detailed answer. I was asking because I didn’t find any search capabilities in the app I used, so I will stick to web version for searches until apps implement it.