Add generator search through Algolia yarn index #763
Conversation
There was a problem hiding this comment.
Hej Tim,
I love it. Thanks for the PR 💖
Our package which is creating the current generator list was completely rewritten a year ago but still has some hickups from time to time. Hope this will be rock solid (stable) solution.
I just wonder if it would be possible to blacklist a few modules like we did before. Must say that I’m not sure about the importance of that blacklisting feature because the blacklist wasn’t updated for quite a while.
Cheers, Michael
P.S. Greetings to Luky and Haroen 😘
|
Glad you like it! The list currently only return packages that have I think this solution might be more stable as you'll not be the only one relying on it, and we actively monitor it for any issue. The disallowed list being pretty short, I think a quick fix would be to not render results that are matching an element of that list directly from the front-end. P.S: Your love has been passed to them :) |
|
Awesome PR thanks a lot for sending it :) I think the ranking algorithm will remove the need for the blacklist. I'm happy to drop it personally. |
app/assets/js/generator.js
Outdated
| apiKey: '14ddf54fd4f3435c1cd4038395a0cf10', | ||
| indexName: 'npm-search', | ||
| searchParameters: { | ||
| filters: 'deprecated:false AND keywords:yeoman-generator' |
There was a problem hiding this comment.
filters: 'deprecated:false AND computedKeywords:yeoman-generator'
I changed the Yarn index to now have a subset of the complete registry, they need the keyword and start with generator-. You can see (and edit if necessary) the code for that here
Don't hesitate to reach out if there's any questions about this
|
I guess it won’t be possible to highlight the generators maintained by the Yeoman team like we used to do?
@SBoudrias I’m not sure how important that is since a lot of them are pretty much unmaintained 🙈 |
|
How can you recognise if it's maintained by the Yeoman team, |
| var name = highlight(item, 'name').replace('generator-', '') | ||
| var url = getUrl(item); | ||
| var description = highlight(item, 'description'); | ||
| var authorName = item.owner.name; |
There was a problem hiding this comment.
var authorName = item.owner.name === "yeoman" ? "The Yeoman Team" : item.owner.name;
that would do what you suggested @mischah, could even add a custom class or whatever if needed
|
It was stored in a Hold on. I think this is obsolete because your solution makes it christal clear the a generator is from the team:
|
|
yeah, already clear enough as-is :) |
|
This change looks great to me :) @mischah I'm good to have this merged in if you are! I'm wondering if we could also update the |
|
Should be possible, but it's not that easy making an interactive cli. How are you fetching the search now? |
|
just be sure that the change I suggested in #763 (comment) is addressed before merging :) |
I just want to check the possibilities to improve the experience on small viewports first.
Via https://github.com/sindresorhus/npm-keyword over here: |
This will only include the packages starting with `generator-`
|
It’s live and it’s looking good 🎉 Thanks, @pixelastic @Haroenv 💝 @SBoudrias I’m going to schedule a tweet from the Yeoman account for Monday to gain more attention. |
|
Thanks @pixelastic 😉 |
|
Hej @SBoudrias, one last note about this change: I’ve just closed all existing issues of yeoman-generator-list and removed the travis cron job. Transformation completed.
|
Hello,
Tim from Algolia here. I wanted to suggest this PR that updates the current generator search to provide a search experience similar to the one you can have on the Yarn website.
You can try it yourself right here.
This actually uses the exact same dataset as the one used on Yarn (containing all the npm packages), but filtered to only show yeoman generators. Results are ranked based on the popularity (number of downloads and dependents). All the data is coming directly from npm, and is updated in real time whenever a package is added/deleted/edited. Others (like Gatsby, Babel or CodeSandbox) are also relying on the same index.
It is slightly different than the current implementation, but it has a few advantages: it is faster and handles highlighting and fuzzy matching out of the box. I'd be happy to discuss the implementation further (there is still room for improvement regarding the display on mobile), but I first wanted to check your interest in such a feature.