RT @kdzwinel: @tomayac AFAIK there is a fastpath in querySel for “#id” and “.bla”, but it looks like it’s still 60% slower than getElById…
My approach usually is to query as specifically as possible to avoid selector parsing; I still use getElementById and getElementByTagName.🶠https://t.co/b4crY8g6UZ
estellevw I get free sushi if 1,000 people take the SpeedPerception challenge speedperception.com/challenge. It’s a #perfMatters cause. Pls RT
kdzwinel There is a colored dot next to each “Product” in the Network panel. I’m even more intrigued now. pic.twitter.com/sNTZunHNJF
@kdzwinel @JonGarbee You generally should get a high score, but 100/100 can still mean slow—PSI score isn’t correlated to latency & is no direct ranking signal).
RT @kdzwinel: We have a winner 🆠@tomayac showed me a gem that already has a perfect 0.
My personal best is 39. I can’t do anything wrong…
@kdzwinel @JonGarbee 🤔The secret is ruleImpact (https://t.co/t5pqoNEYRt), an open-ended score of the relative impact of violating the rule.
@kdzwinel Final tip: things add up. https://t.co/x6vaDGioxy https://t.co/kCFAuS0ahP
@kdzwinel First, slow down your server and make it respond only after 3s or so. Second, deliver an empty page and load all content via XHR.
@kdzwinel At least GTM makes it impossible to add render-blocking scripts… I see Optimizely on many customer pages—and it’s a perf killer :(
@kdzwinel May I suggest you add a synchronously loaded A/B test? https://t.co/LeL7bv6szw