Nice! Looking forward to it. Trying to migrate my teamβs workflows to continuous tasks/TUI for day to day dev and this got raised as enough friction to stick current workflow of separate terminal tabs/windows
Nice! Looking forward to it. Trying to migrate my teamβs workflows to continuous tasks/TUI for day to day dev and this got raised as enough friction to stick current workflow of separate terminal tabs/windows
@juri.dev is there some way to search the logs for an @nx.dev continuous task, aside from copy and pasting the entire TUI output into a separate file and searching? Or, is there a way to tail the output for a task(s) to a file? There's the MCP server, but is there a simpler way to see/query logs?
π₯²
...was counting on being auto-memoized for performance reasons. An innocent change blew up the performance of the whole tree. Handler functions created in the parent that had previously been memoized now changed every render, messing up the memoization of the child components, etc. etc.
I think at this point my team and I definitely want warnings, either via eslint or at compile time, when a component won't be memoized.
In the real-world situation I'm pulling that sort example from, it was a big, ~1k-line component that renders a couple costly sub components, each of which...
Same issue though when copying and using sort playground.react.dev#N4Igzg9grgTg...
It's so silent and surprising that it has me considering whether my guidance to my team should be to stop using useMemo altogether, just use IIFEs and let the Compiler do any and all memoizations
Will this help situations where the Compiler bails out on certain uses of useMemo? I've been surprised a few times in the past couple weeks where adding a useMemo that follows all rules and looks harmless accidentally de-optimized a component playground.react.dev#N4Igzg9grgTg...
Got to the bottom of it π github.com/facebook/rea...
We keep having to 'use no memo' in components that use @tanstack/react-table, though. I think there's something with the useReactTable hook that is relying on over-rendering/breaking some rule of React. Haven't gotten to the bottom of that
I enabled it February on a very large Next.js project I joined that had little to no memoization. Re-renders were so bad taking a profile would fail most of the time. It worked wonders! Almost no one on the team knew about it! I was very surprised. Next.js making it an "experiment" doesn't help
"Masked security forces captured an opposition leader in the coastal province of New Jersey as regime officials continue their months-long crack-down against pro-democracy elements in the embattled United States. The arrest comes as the regime issued calls for summary detentions without trial."
I found and fixed a bug in @nextjs.org! π€ #OSS github.com/vercel/next....
@arktype.io when's the new package getting published? π¦ π github.com/arktypeio/ar...
Video of the international student at Tufts being arrested by "federal authorities" in Massachusetts has been released and it's terrifying.
They're not even uniformed officers. Just secret police thugs in hoodies and masks.
From WCVB: youtu.be/PuFIs7OkzYY
@juri.dev I'm curious, with the TS Go future how has that changed the @nx.dev team's thinking for TS monorepo setups? You think package manager + references setup is still the best default, or single project + path aliases and adopt references as needed?
Just ran into the fact that you can't schedule send text messages from iPhone to Android, only to other iPhone numbers π€¦ββοΈ What a fucking joke
The Department of Government Efficiency isn't interested in efficiency. Or transparency, or cutting costs. It's real goal is to destroy the federal civil service. In other words, regime change. www.theatlantic.com/ideas/archiv...
Just saw a farmer say, "you don't have to admit you were wrong, you have to admit you were lied to, " and idk that seems like a good framing to call people in
Also how can I get suggested imports working again? That functionality seems to be broken in VSCode with the references+workspaces setup, using either pnpm or npm. Having to manually write out import paths is unfortunate. Is that a known issue?
Looks like the esbuild, swc and rollup executors are all set to not allow generatePackageJson github.com/search?q=rep...
Maybe this is coming in Nx v21? π€ This is a hard blocker for being able to adopt a references+workspaces setup.
@juri.dev another question when using the new @nx.dev TS setup. I have Node.js apps built using the esbuild executor, using the generatePackageJson option to install only that app's deps in prod.
With the new setup there's an error about that option being unsupported π«€ What's a viable alternative?
@juri.dev, I just finished doing a deep dive on all the latest Nx & TS work. Really exciting stuff!
When following the new suggestion to use references+workspaces, is the withNx bit for configuring a Next.js app nx.dev/recipes/next... even needed anymore?
Windows?
Before and after! www.descioli.com/grove
Love this!
I just tried this out on my site and caught some accidental enlargement in places other than the first paragraph, e.g. an article where I had list items in a ul. Solved it by using `p:first-of-type` instead of `:first-child` π
Yup, Iβm tracking the same threads about it. Maybe this will be the year reading from a big context is performant instead of a pitfall! π« π€
I haven't put the compiler through its paces on these cases yet, though, so I could be happily wrong!