Kevin Farrugia

683 posts

Kevin Farrugia banner
Kevin Farrugia

Kevin Farrugia

@imkevdev

Web performance consultant. Be kind. He/him. Find me on BlueSky: https://t.co/S4IV2Tztn3

Malta Katılım Ağustos 2010
539 Takip Edilen262 Takipçiler
Rich Harris
Rich Harris@Rich_Harris·
@nargilelelele i suspect this is mostly a statistical artifact — there are likely very few client-rendered astro apps
English
1
0
1
194
Rich Harris
Rich Harris@Rich_Harris·
A few takeaways from this remarkable data: 1. Yes, SSR matters. SvelteKit users sometimes get frustrated that it's too SSR-centric, and we have plans to address that in the next major version. But SSR is the default for a reason! Our ultimate responsibility is to end users
Ben McCann@BenjaminMcCann

Some data from Google (big thank you to @rick_viscomi and @imkevdev!) Client-side-only SPAs are extremely detrimental to latency as they add a network round-trip before rendering begins. You can win at LCP by making it harder to build client-side apps😉

English
21
21
337
100.3K
Kevin Farrugia
Kevin Farrugia@imkevdev·
@samthor @Indiequest1 @malchata However I cannot tell the extent to which it affects Kuto. I do not expect it to have a significant impact and from the tests I ran so far, the impact is tiny.
English
0
0
0
31
Kevin Farrugia
Kevin Farrugia@imkevdev·
@samthor @Indiequest1 @malchata The core problem is that when streaming, V8 "guesses" whether the script will be used as a module or not, since that will affect how the script is parsed. If it guesses incorrectly, it ignores the streaming parsing results and reparses it again.
English
1
0
0
36
Kevin Farrugia
Kevin Farrugia@imkevdev·
@firt Is there a list of affected APIs? We noticed that the WebShare API is not working well either (even when the app is opened in Safari browser)
English
0
0
0
66
Ivan Akulov
Ivan Akulov@iamakulov·
Life update: after ~6 years of doing web perf consulting, I turned the page and joined @framer. I’ll be leading site performance efforts there, together with a team of amazing engineers 🖤
Ivan Akulov tweet media
English
26
2
233
15.4K
Kevin Farrugia retweetledi
Stoyan Stefanov 🪩
Stoyan Stefanov 🪩@stoyanstefanov·
🪩 Web Performance calendar bonus 4 🔗 calendar.perfplanet.com/2023/server-ti… 🎇 Nishu Goel (@TheNishuGoel) not only shows us 70% perf improvement with the help of Server-Timing 🛠️ but also ships a new open source tool to go with the article And that's a wrap. Happy 123123 day and 2024 year!
English
0
6
26
4.7K
Kevin Farrugia retweetledi
Firefox Nightly 🔥
Firefox Nightly 🔥@FirefoxNightly·
<iframe loading=lazy> has landed. Lazily loaded iframes are only loaded when visible, speeding up initial page load. See bugzilla.mozilla.org/1860729
English
8
114
460
236.6K
Tim Vereecke
Tim Vereecke@TimVereecke·
My Top 3 of missing Performance features: - DPR capping - Vary based on a specific cookie - CSS property to render elements only when closed #webperf
English
2
2
14
1.2K
Kevin Farrugia retweetledi
Addy Osmani
Addy Osmani@addyosmani·
Tip: In @ChromeDevTools you can now hide network requests from Chrome extensions! 😍 Huge for focusing on just your code when profiling performance.
Addy Osmani tweet media
English
28
240
1.4K
164K
Kevin Farrugia
Kevin Farrugia@imkevdev·
@iamakulov This was client side only, no server side rendering. 🫣 And similarly, I wouldn't expect the batching to have such an impact. Would the "yielding" in performUnitOfWork work even when not using useTransition? I will try to create a repro somehow.
English
1
0
1
75
Ivan Akulov
Ivan Akulov@iamakulov·
@imkevdev But also any chance the client was already using <Suspense> anywhere in the app (eg for React.lazy) + was doing SSR/SSG? <Suspense> boundaries get hydrated differently with React 18; that improves TBT a lot and should improve with INP too
English
1
0
1
79
Kevin Farrugia
Kevin Farrugia@imkevdev·
@iamakulov have you come across any case studies on the impact to Long Tasks or INP after upgrading to React 18 (before even using the new APIs)? TIA. 🙏
English
1
0
0
175