Brendan Kenny

@patmeenan @tunetheweb @SimmerVigor @hnasr To derail further: there's also the fun that's been had with *uncompressed* PNGs but serving them with brotli and seeing improvements. Like thejh.net/written-stuff/… and twitter.com/erikcorry/stat…

Thanks to @tunetheweb for demystifying browser web performance for us, fantastic blog on image decoding (didn’t even know there was an attribute for that?) check it out. We often know what happens between A to C right after tunetheweb.com/blog/what-does…

@Lakbychance @tunetheweb @____lighthouse I added an explanation on the removal of TTI to #scoring-changes" target="_blank" rel="nofollow noopener">developer.chrome.com/blog/lighthous…, but FWIW I think you already have the case against TTI exactly right :)

@tunetheweb @____lighthouse @brendankenny And yes TBT does factor in the “long tasks” bit of “TTI” already so lazy loading main thread intensive below the fold resources (image comes to mind) should reflect in TBT calculation as well.

Lighthouse 10.0 is here!💡🚢 ➡️updated scoring (we'll kind of miss you TTI) ➡️bfcache ➡️complete type declarations ➡️read more: developer.chrome.com/blog/lighthous…

@GrahamTheDev @____lighthouse Yeah, there should have been more detail in the post. I just added some explanation of TTI's removal in #scoring-changes" target="_blank" rel="nofollow noopener">developer.chrome.com/blog/lighthous…. Let me know if that helps make it clearer.

@____lighthouse Is there a blog somewhere explaining why TTI was removed?

@patmeenan @tunetheweb @juan_ferreras_ @RealWebPageTest @HTTPArchive It's enabled by default so should be good to go for the March HTTP Archive run. It appears that there are sometimes issues with running it with --headless (you get an extra failure reason about using window․open()) that's being looked into

I'm particularly excited about the bfcache audit added in Lighthouse 10. There was already a great testing tool in Chrome Dev Tools (#test-to-ensure-your-pages-are-cacheable" target="_blank" rel="nofollow noopener">web.dev/bfcache/#test-…), but having it here as well opens up lots more opportunities for making this important optimisation more widely known.

@addyosmani @RReverser @Paul_Kinlan I can't find the tweet I learned it from now, but you can sometimes get it to continue. Something like "Your answer got cut off. Continue your response starting with 'To save...'"

@RReverser @Paul_Kinlan Thanks for the tip! I’ll give it a few more tries.

I asked GPT Chat to describe how to make a Todo list app in vanilla JS. And it did! jsbin.com/patihar Simple example, but it's incredible.

@torturecrush going to try this person's squish method next year: youtu.be/FbM7Y4-4iDs?t=…

@torturecrush the articles out there are *so* useless. youtube is much better for demos (caveat: I was only semi-successful as a first time grower this year). I used this method: youtu.be/2kjOnnhcLfs and like his, they could have used longer, but still good in the center

she got so big the plastic trellis was buckling so now she gets her own suite at the plant hotel also, I managed to get this watermelon plant to bloom again so not pictured are 2-4 more growing watermelon babies. maybe I’ll go snap them now, love a morning garden stroll

@hankchizljaw From your description and running the demo, it really is a layout shift that's happening. Are you asking if this technique is possible to do without a layout shift, or if this shouldn't be penalized because it's a tradeoff for progressive enhancement?

@ryanflorence Just remember diagnostics don't impact the Lighthouse perf score, they're there to help break down what's going on with a metric. So if delayed font display was dragging down FCP, the only penalty is the measured FCP delay.

@SeaRyanC Consider not reading more recent posts, however.

@SeaRyanC acko.net/blog/how-to-fo…

What was the website that explained the Mandelbrot set by starting with how addition, multiplication, etc are transforms on the fabric of the complex plane, constructing Julia sets from that, and so on, using tons of animations?

@cramforce Not in the type system, no. For literals you can enforce the type is a literal, but there's no way to check the value was declared as one

People who know typescript better than me: Can I express in the type system that function parameters must be root level declared const or literals?

@tom_forsyth Stumbled across it a few years ago and (as a Texan) was *very* surprised. Somewhat unique flavor profile as well, would be interested in what you'd compare it to. Drama behind the scenes: nytimes.com/2014/12/28/bus…

It's a single malt whisky - they even spelled it right - but it's from TEXAS!?!?!? I mean how good can it be? Er... actually... it's quite nice indeed. Jesus Texas. You even managed to fuck up fucking up. Good job. Or bad job. I'm very confused.

Balcones single malt whisky. Gaze upon this abhorrence!

@SeaRyanC Interesting! How does github.com/microsoft/Type… fit into that?

TypeScript tracks every type with a unique ID, which is generated sequentially. Sometimes a bug manifests where the relative ordering of the ID matters (it never should), which is a pain to figure out. If someday you find an undocumented --backwardsDay flag in tsc, that'll be why

@j_holtslander @ChromiumDev otherwise do you mind filling out github.com/GoogleChrome/l… and we can do a more in-depth attempt to reproduce than just a trying it out on a spare test phone?

@j_holtslander @ChromiumDev Yeah, that's odd. Maybe it's the in-app browser? Content blocker? Do you see the same thing opening pagespeed.web.dev in the default mobile Safari? The text input is baked into the served HTML so doesn't depend on JS to appear

Go test out the new PageSpeed Insights, live now! pagespeed.web.dev 🚀 Details on the changes: web.dev/whats-new-page…

@isAdrisal @____lighthouse @addyosmani @ChromeDevTools Thanks for catching this and reporting! I have a fix for it in github.com/GoogleChrome/w…, will hopefully land soon.

Hey @____lighthouse, web.dev/measure now provides desktop results rather than mobile with no way to toggle between them which feels like a regression. Are there plans to switch back to mobile results? @addyosmani @ChromeDevTools

@rbuckton @typescript @drosenwasser Yes! Wasn't sure the reason for that burst of jsdoc PRs, but was great to see them landing :)

@brendankenny @typescript @drosenwasser Done forget this one: Type Parameter defaults in JSDoc @ template tags. github.com/microsoft/Type…

Our release candidate for TypeScript 4.5 is out, and is the best way you can help shape the release! Read up on what's new and what's changed since the beta, and try it today! devblogs.microsoft.com/typescript/ann…