I haven't released any personal projects for five years. I've built plenty of prototypes, but none of them ever felt ready. Here's what I've been doing, and what I'm going to do next:

btw the areas where you can start scrolling or resizing here are the same we wait until you start dragging to detect your intention, based on whether the cursor moved more horizontally or vertically

it’s often possible to structure your UI so that the dynamic part of the content doesn’t affect the cumulative layout when that isn’t possible, it’s best to target being correct for the most common case

anything we can do to make the dashboard better?

I’ll forward this to our o11y team, thank you for reporting!

we miss you 🖤

just the page root and we’re only doing this because we have a full-height sidebar. I wouldn’t recommend doing this on all websites

we started with an (any-pointer: coarse) query and changed it because of feedback from folks using Windows laptops w/ touchscreens 😅 limiting to the y axis is v good though, and imo you should only do this when you have full-height sidebars or other specific layouts

pointer media queries are evaluated against the primary pointing device, so it will still match (pointer: fine) in that case developer.mozilla.org/en-US/…

we made sure pull to refresh still works 🙂

we do this with overscroll-behavior: none but we gate it behind a (pointer: fine) media query, so that it doesn’t impact touch devices. this means you can still pull to refresh on mobile

thank you for the feedback! we’re working on improving the navigation on both desktop and mobile for the small UI, lack of animations, etc. can you provide specific examples?

oh wait. we actually already have a setting for this if you go to “My Notifications” it’s “Import detection”

thank you for the feedback! I’ll share this with the team working on it

heard! we’ll improve this

thank you for building an awesome library! :)

yep! this is the second place we’re using it in the dashboard

you can try out these filters on the deployment list pages in our dashboard for more details on how we engineered them, see this thread

you can try out these filters on the deployment list pages in our dashboard for more details on how we designed them, see this thread

when users type a query that we can't find a simple match for, we switch to determining the filters using AI this is primarily done by: • generating a zod schema based on the relevant filter definitions • generating an object that matches that schema based on the user's query but there is additional nuance, like: • select/multi-select values are matched to their closest valid value on the server • the LLM can include a special sequence to indicate that a value should match the current user • dates are passed back raw to the client so that they can be parsed with the user's local timezone

5. natural language processing this feature still relies on the same filter definitions as everything else in this thread those definitions are isomorphic, so we can import them directly in the client-side code, but also in the API route for this feature