patmeenan.com
Web performance geek on the Chrome team at Google.
Formerly WebPageTest, Meta, Cloudflare, AOL...
59 posts
1,914 followers
161 following
Regular Contributor
Active Commenter
comment in response to
post
Not an option that came up in the new device setup flow. All paths ended with trying to text a code to the phone number. Only other path would have been a 3-5 day wait to link it to a different number.
comment in response to
post
Before it comes up, sending the code to the email for the account is theoretically an option but it follows that with a need to verify by texting a code....
Feels a lot like the teams working on it live in a bubble where not having multiple iOS devices isn't conceivable.
comment in response to
post
It looks like you have to go REALLY far back for a browser that doesn't support "preconnect".
At this point I'd probably just recommend picking whichever is more appropriate (depending on the liklihood to need to fetch a resource not in cache).
developer.mozilla.org/en-US/docs/W...
comment in response to
post
Seriously doubt it. AFAIK, they hoover everything they can find. Seeding a bunch of code with fake accounts on GitHub would probably be pretty effective.
Particularly if you're a nation-state targeting specific niche things or algorithms.
comment in response to
post
Kind of like the next iteration of "why we can't have nice things" (yeah, I miss the web from pre-2000 when it was still relatively innocent).
comment in response to
post
Just saw abug come across the issue tracker last week where someone was using utf.
Feeling like exposing the status text to node has a lot of rough edges.
comment in response to
post
No "Apple Intelligence" on 13's (I'm on the same). 15+ if I remember correctly.
Just regular "actually works" non-AI 😂
comment in response to
post
FWIW, it's not even necessarily "older" devices. Just about any current device with a MediaTek SOC is still just as slow (I use recent Moto Play devices for local testing).
They're cheap and have good specs "on paper" (core count, frequency) but are missing key architectural features.
comment in response to
post
Thanks. I think the right people are giving it the right level of importance at this point - it's just going to take a while to reach critical mass and be at the point where it is one-click to enable.
CDN's, infrastructure teams and large sites are all actively working on it.
comment in response to
post
It can be. Unless you are the one sending the animated meme's, trying to figure out why they are pasting as static images 😂
comment in response to
post
There were some unintended side-effects when I had it turned on. Most notably, some animated GIF's would not play (maybe in slack, not sure if also in messenger) which was really confusing.
comment in response to
post
Last I knew there were some concerns about trusting the connection itself as the authority (just the cert) but maybe those concerns were overblown.
Not sure if there are any plans to enable it for HTTP/2 as well.
comment in response to
post
Thank you. That means a lot to me coming from y'all. The performance community has been an awesome group of people for so many years.
comment in response to
post
Where did it take you?
As far as I can tell, quantum mostly puts public-key at risk because of the factoring but symmetric (password at rest) only reduces the effective bits by half and is basically safe.
For same protection double password length and AES bits (eg 256 to 512).
comment in response to
post
Settings -> VPN & Device Management
Click on any unused device management profiles and remove them (assuming it is coming from a managed profile)?
comment in response to
post
Sure. You use link rel=content-dictionary to load a dictionary that can be applied to document requests or fetches.
See Etsy.patrickmeenan.com to see it live for the page html.
No different from SW where the very first load can't use it but everything after can.
comment in response to
post
Now that compression dictionaries are available, they're probably a cleaner and easier solution than stream stitching. The common bits can go in the dictionary and just compress away (Chromium-only right now though).
comment in response to
post
Example ($399 instead of $1k): humanspine.com/steelcase-le...
comment in response to
post
Corp discount FTW. Should be able to find it for far less than retail.
comment in response to
post
+1 for the leap. Far better than the HM chair I had before.
comment in response to
post
Until then, I have a reference Cloudflare worker that can do it all with a bit of config: github.com/pmeenan/dict...
Still need to document it better though.
comment in response to
post
It's remarkably simple to do it all on the edge (delta compression of static assets and stand-alone dictionaries for dynamic content).
I expect broad adoption will ramp more heavily once CDNs expose the capability.
comment in response to
post
I expect using it for e-commerce html will have the biggest impact. They tend to have a decent number of pages within a session and can reduce the transfer size by 50%+.
You can usually get all of the above-the-fold content into the first packet.
Also easy to draw a direct line to revenue.