Barry Pollard

Interested in how the modern web works? Think #webperf is important? Then check out my book “HTTP/2 in Action” with @ManningBooks at manning.com/books/http2-in… And if you’ve ever considered writing a book then check out the story behind the book at tunetheweb.com/blog/writing-a… #http2

@xopek59 FYI, we (Chrome) are changing the values of this and this will start to be rolled out to users from next week: chromestatus.com/feature/633037… If you are blocking based on values of 16 or 32, then this will block real users going forward!

The navigator.deviceMemory API looks trivial, but it can reveal a lot about bots. Inconsistencies may show up when attackers try to spoof it, making it a useful fingerprinting signal. Full post: blog.castle.io/deep-dive-how-… #botdetection #infosec

Alright, I wrote it. domenic.me/streams-standa…

@domenic My vote is for you to get pulled back into that world… we miss ya! And especially your guidance.

Very tempted to write up a response to blog.cloudflare.com/a-better-web-s…... It has some good points but a lot of questionable ones and some that are just confused and wrong. But I'm not sure I want to get pulled back into that world.

Don't suppose anyone in my circle knows anyone working at X engineering? Need to discuss a potential issue preventing some users from Chrome 146 logging in. @XDevelopers @XEng

Is there anything more frustrating in the world that trying to create a child's @EA games account? Inconsistent login options (Google sometimes appears, sometimes doesn't), you need a desktop app not a mobile app, you can't login using EA ID online, but can't set an email. Sigh…

@NigelWhitworth You have both `height` and `width` set to `auto`, meaning the browser doesn't know how to use those supplied dimensions. You only need to use `auto` if setting a max-width (or max-height) and should only set `auto` to whichever you aren't setting a max value for.

@tunetheweb Hi Barry, can you tell me why it says "Unsized Image Element" when I have specified the height and width please. example page : universeguide.com/fact/multiverse

@iSaumya Sure do that. Will take a look when I get a chance!

@tunetheweb This is not a widespread issue. I have seen this issue on a few sites and ensured that the pages are not speculated before. If you want, I can DM you a screen recording and HAR. I've looked at everything, but this makes no sense. Same outcome whether I use prefetch or prerender

Hi @tunetheweb, I've recently come across a weird edge case issue with the speculation rule. That is when a page is prefetched or pre-rendered, the cookies are not passed, causing the page content to be different. It's sporadic and not easily reproducible. Any idea how to solve?

@iSaumya I’m not aware of a general issue where cookies are randomly not sent. And I’ve helped a lot of people implement Speculation Rules on a lot of sites. So something specific must be happening that triggers this for your site.

@iSaumya I would suggest: 1) Reading that doc in full. 2) Trying to figure out how/why this is happening. It may seem random but there will be a reason. 3) Creating a minimal reproducible test case so we can investigate further. Unfortunately without that, it’s difficult to help further.

@iSaumya No, not necessarily. I’m saying prefetch usually has less consequences than prerender, as JavaScript doesn’t execute (e.g. if basket count is fetched dynamically from client-side API call). If the basket count is in the HTML then both prefetch and prerender can have this issue.

@tunetheweb Also, you recommend using prefetch than preload to mitigate this out-of-date issue? Cause in the doc I'm not sure there is an easy global way to fresh prerender on each hover. As that's what should happen.

@iSaumya And can you explain what you mean by the cookies being missing? They aren’t there in the prerender HTML request in the Network tab in DevTools? Or they are no longer in the cookie store in the Application tab in DevTools?

@iSaumya Of if the page with 0 basket is cached from previous visit? Do you allow caching of your HTML pages? (I’m not that familiar with Woocommerce to be honest but is the basket count part of the HTML? Or fetched by an API call on page load?)

@jaffathecake @ganyicz @Una Curious if this is another instance of this bug that was recently fixed by @yoavweiss : issues.chromium.org/issues/4591356… Can you retest in Chrome Canary which has the bug fix and let us know if you still see it there? If so a crbug.com with minimal test case would be great!

@ganyicz Please file this at crbug.com (cc @Una)

This css unit DESTROYED our performance ☠️

Web Performance Calendar day 23 with @tunetheweb on configuring caches to ignore non-content-affecting URL params and reclaim performance. No more of this query string = cache miss nonsense. calendar.perfplanet.com/2025/fixing-th…

@NigelWhitworth @Google @JohnMu Usually means there's not enough Android Chrome visits to cross our publishing threshold (maybe you have a lot of iOS traffic that doesn't get measured in CrUX?), but you do get sufficient Chrome desktop visits. More details on eligibility critieria here: #eligibility" target="_blank" rel="nofollow noopener">developer.chrome.com/docs/crux/meth…

How can one of my site's CWV pages disappear on Mobile but still exist on desktop? Does anyone have this trouble? Site is responsive so showing same page to both types. @tunetheweb @Google @Johnmu #CWV

@iamakulov If you measure time between rAFs you’ll likely see 60Hz a lot of the time. Even on monitors that support more. Thats my experience anyway.

@iamakulov I’ve not been able to find definite info so I hand waved it there but some issues: - What monitor uses (e.g. Apple’s Promotion dynamically adapts) - Multi-monitor setups - Main thread and compositor may rate be different. - rAF may be throttled TLDR: it’s complicated

So @ryanflorence had a good chat with @mmocny on this yesterday, but for the rest of you out there here's some things to know. INP measures (as much as possible) from the moment you interact (click a button, type a key...etc) until the pixels actually update. 🧵 1/12