rschristian.dev
Implicit "It depends" | Open Sourced Shenanigans http://github.com/rschristian | @preactjs.com stuffs
612 posts
988 followers
98 following
Regular Contributor
Active Commenter
comment in response to
post
I'm with you in that it's worthwhile so long as it benefits the end user experience (admittedly, I'm far less concerned w/ DX than most), though at some point you'd have to look at it like it's lipstick on a pig -- a minor improvement that doesn't attempt to address the underlying issue.
comment in response to
post
Supposedly that's because it's being killed off :(
Was lucky enough that it worked for my needs this past filing year, was an absolute breeze & by far the quickest I've ever gotten it done. I really, really hope it sticks around or at least comes back in a few years.
bsky.app/profile/rsch...
comment in response to
post
Inb4: Something something, should teach both sides of the holocaust
comment in response to
post
The client router swaps the entire page body? Huh, wouldn't have guessed that. Fair enough though!
comment in response to
post
What's the problem with that? Haven't yet ran into anything myself (or don't remember if I have), I'm curious as to what you've seen.
comment in response to
post
Not that it's that much better, but at least Bezos, Zuck, and Cook could reasonably be sat across from the President when talking about policy as they undeniably head massive American companies. Paying lip service makes some sense.
Vercel on the other land? lol
comment in response to
post
When reading this article, a quote came to mind: "Those who enjoy brutal honesty are far more fascinated by the brutality than the honesty"
This rings true here, as well. Not an ounce of this article is heartfelt; it takes far too much joy in its derision and contempt.
comment in response to
post
I'm not sure that was their intended meaning, although it's a bit ambiguous. I interpreted it instead as pointing out the hypocrisy in personal vs corporate actions and how the courts choose to enforce laws.
comment in response to
post
And Preact has docs to showcase how to use it in that sort of setup: preactjs.com/guide/v10/no...
Rest of the ecosystem though... can certainly be a bit rough. Many tools rely on bundlers/treeshaking to function.
comment in response to
post
If anyone does see folks saying things along those lines in favor of Preact, please let us (the team) know. We'd be quite happy to step in and ask that they don't do that, we don't want or need people evangelizing us and putting down other tools or maintainers who simply choose other tools.
comment in response to
post
I think internal gets a pass, if for no other reason than that I ask far worse questions internally on a daily basis and I really, really don't want those to count for anything.
comment in response to
post
Frameworks are just tools in a toolbox, they're nothing to evangelize or fall into tribe mentality over and I think every maintainer would say the same.
I for one have written more web components in the past ~3 months than Preact components and it's been a pretty nice experience!
comment in response to
post
Er, no? That's not what the sibling selector addresses.
The point of TW here (or some of the approaches mentioned above) is that all CSS becomes highly focused, there is no doubt upon where/when it's used. Exceptions (as this still would be plain CSS) become commented & documented usually.
comment in response to
post
IMO, yes, though there’s other approaches mentioned above that work well too.
comment in response to
post
That’s such an awesome approach
comment in response to
post
Especially not when JS/conditional rendering gets involved.
i.e., `img + p { … }` — there’s no easy way to determine whether or not this is used anywhere on your site, becoming impractical to change at all once you’ve hit a few dozen pages. Visual regression testing might help a few but not all.
comment in response to
post
Not app-wide, not accurately.
comment in response to
post
Restricting yourself to those helps and mostly solves the issue, fully agree.
Problem is the global stuff that’s not tied to any component or particular bit of the UI. Every project has some of that and, TW usually helps keep that at a bare minimum for the undisciplined (like myself, admittedly).
comment in response to
post
Really? Append-only style sheets are definitely the norm in most code bases I've worked in unfortunately, everyone's too scared to touch anything as the effect of the change is a complete unknown.
CSS modules and/or scoping (be it manual or new built-in) certainly goes a long way though.
comment in response to
post
Interesting, have you never struggled with refactoring plain CSS before?
Admittedly that's the biggest selling point of TW to me, plain CSS can often be near impossible to predict. The global nature of it means removing a style might break an unexpected & unrelated portion of the UI.
comment in response to
post
It's amazing that I've been on iOS for 12 or 13 years now and this hasn't noticeably changed over the years, it's still as broken and unreliable as it always was.
Only, as time has gone on, more and more companies have lobotomized their mobile web experience which sucks :/
comment in response to
post
Sales of a the mini absolutely cannot be used to claim “everyone just got used to bigger phones”, device choice doesn’t happen in isolation and is far more complex than that.
comment in response to
post
I don't think we got used to them so much as begrudgingly deal with it as there's very few alternatives on the market, or at least that's my excuse for owning one.
I hate that I can't easily use my phone one handed, would trade it in a heartbeat.
comment in response to
post
Yuuup, see that all the time + accessing the linked module's devDeps.
Resolving modules through the linked file location is such an obvious & awful footgun, I really don't understand why that utility was ever shipped. There's so few situations in which it can't go really (confusingly) wrong.
comment in response to
post
So, so much better, avoids the many pesky & completely non-representative issues that linking brings.
I think I recommend it at least once a week to someone.
comment in response to
post
Prior to ES5, there was a legitimate issue in that `props.class` was not allowed, but that's very much not an issue today and only destructuring can present an issue (though easily avoided by using "clss", "cls", etc., if needed).
I think they simply don't want to risk breakages by allowing either.
comment in response to
post
That's the "deep design and creative responsibilities" that Jony has assumed and seemingly immediately dropped, or do we just not understand the vision here?
comment in response to
post
The math on that disgusted me tbh
comment in response to
post
Not sure how well it works but I believe it's just a matter of building w/ a different target: parceljs.org/features/tar...
Shouldn't be all that different from creating an SSR build w/ Webpack, Rollup, Vite, etc., I imagine.
comment in response to
post
I first discovered Preact during a university project 5+ years ago and it's been cool to see it grow and mature since then. Hope to see it continue to grow too!
Bonus: It hasn't need a breaking change this whole time. 200k -> 6m downloads per week in a single major version over 5.5 years.
comment in response to
post
Yup -- I remember years ago running into issues when a new TS version got released and the yarn team hadn't yet had time to patch it. Was quite uncomfortable with that once I learned of it.
comment in response to
post
npx doesn't count as a download for the stats? Got any reference for that?
That'd sound pretty odd as the npm repo looks to track downloads rather than installs (and npx basically just installs into a temp dir if it needs to anyhow).