I've absorbed so much new information about Meilisearch this week, and I still feel like there is so much I don't quite understand. So today's TIL is filled with things I wasn't sure about... but I did learn things.
@wouterj cool work! I recently started implementing an "autocomplete/autosuggestion" helper for the SF console, backed up by #meilisearch , and would like to know how you implemented the "right scrollbar" :)
Is there a public repo for your example?
Or are there any other SF console helpers that you know?
Anyeay: thank you again for your great work on SF!
I've just pushed v1.1.1 of the mongodb_meilisearch #Ruby gem. This fixes a bug that new users are likely to encounter where a new index wasn't having its filterable or sortable attributes autoconfigured.
I also updated some JSON keys used behind the scenes in accordance with deprecation warnings from meilisearch-ruby
If anyone's using my mongodb_meilisearch #Ruby gem , i just noticed a bug where it's indexing additional fields it shouldn't. Won't negatively effect your searches, but will result in wasted resources. I'll try and fix that this weekend.
TIL: Manticore has no "stop words" loaded by default. Kinda 🤯 to me. I'd expect it to ask you what language(s) you were planning to index and pre-load them, or offer to pre-load them. Or just pre-load them in its default language without asking.
I've spent the past 2 1/2 days down a rabbit hole of refactoring.
There's no good #Ruby gem that makes it trivial to integrate your #MongoDB backed models with #Meilisearch
I was tempted to just bundle up the files I've been using for > 1month, but i have standards and don't want to screw anyone with something that wastes their time because it only "Works for me"
14 hours later, I've got a radically better codebase, packaged up as a gem, with tests for most of the major stuff.
Managed to get #meilisearch working with my #akkoma instance. Before I tried Zincsearch via the Elasticsearch support, but I couldn't quite figure out why it wouldn't index properly.
I'll test it a bit and improve my #podman setup and then I can share some guide. :blobcatnerd:
I spent most of yesterday trying to fix the mongoid support in meilisearch-rails. After spending nearly a day with it i'd still barely wrapped my head around all the complexity. Lots of ong complicated untested methods.
Just before going to bed my brain realized "this is just a really complicated implementation of a simple wrapper"
This morning, after walking the dogs, I cranked out a little under 200 lines of code that implements basically all of it in a bunch of small, testable, easy to reason about methods.
Haven't written tests yet, but this afternoon I actually used the new lib and have working, configurable, easy to use, #Meilisearch integration for #Mongoid backed models.
3/3
I intend to live with my new #Meilisearch library for a few weeks and then package it up as a gem for others.
In theory it should be pretty easy to support a relational db too. The only thing it complicates is the unit testing because now you need to connect to mongo AND a relational db.
🤔 not sure if there's enough demand to bother though.