Ryan Townsend
@twnsnd.com
Technology Leader & Distinguished Software Engineer • 20+ YOE in Tech, 10+ as CTO • Public Speaker • LessonsofaCTO.com
So you don't mind eating Meatloaf... the person?
a cartoon character without a shirt is looking at something
Alt: Fry from Futurama narrowing his eyes
media.tenor.com
January 7, 2026 at 3:24 PM
So you don't mind eating Meatloaf... the person?
Distinct lack of capitalisation there!
fry from futurama is shown in a close up
Alt: fry from futurama is shown in a close up
media.tenor.com
January 7, 2026 at 3:09 PM
Distinct lack of capitalisation there!
🤔 @zachleat.com how do you feel about Meatloaf?
January 7, 2026 at 12:55 PM
🤔 @zachleat.com how do you feel about Meatloaf?
lessonsofacto.com
January 7, 2026 at 9:35 AM
Vue for the X-axis, React for the Y-axis and Angular for the Z-axis.
Never forget to center-align in three dimensions!
Never forget to center-align in three dimensions!
January 4, 2026 at 1:34 PM
Vue for the X-axis, React for the Y-axis and Angular for the Z-axis.
Never forget to center-align in three dimensions!
Never forget to center-align in three dimensions!
So if you care about the money then you should care just as much about the data outside conversions too.
January 2, 2026 at 2:45 PM
So if you care about the money then you should care just as much about the data outside conversions too.
I feel my point still stands here: eCom Directors basing their decisions on data from just those who check out are ignoring a tonne of valid data, not just in web perf but more broadly.
Especially given this usually amounts of session-level data, potentially missing much of the research phase.
Especially given this usually amounts of session-level data, potentially missing much of the research phase.
January 2, 2026 at 2:44 PM
I feel my point still stands here: eCom Directors basing their decisions on data from just those who check out are ignoring a tonne of valid data, not just in web perf but more broadly.
Especially given this usually amounts of session-level data, potentially missing much of the research phase.
Especially given this usually amounts of session-level data, potentially missing much of the research phase.
Totally fair and I agree.
When you’re deciding an architecture you need to scope that decision to the current and future users of that architecture.
When you’re deciding an architecture you need to scope that decision to the current and future users of that architecture.
January 2, 2026 at 2:39 PM
Totally fair and I agree.
When you’re deciding an architecture you need to scope that decision to the current and future users of that architecture.
When you’re deciding an architecture you need to scope that decision to the current and future users of that architecture.
In the context of @infrequently.org’s post, I’d argue we always need to look at the whole dataset.
We can’t exclude the non-customers when looking at the cost of admission.
Everyone pays it, regardless of whether they check out.
We can’t exclude the non-customers when looking at the cost of admission.
Everyone pays it, regardless of whether they check out.
January 2, 2026 at 10:34 AM
In the context of @infrequently.org’s post, I’d argue we always need to look at the whole dataset.
We can’t exclude the non-customers when looking at the cost of admission.
Everyone pays it, regardless of whether they check out.
We can’t exclude the non-customers when looking at the cost of admission.
Everyone pays it, regardless of whether they check out.
Reminds me of a client I had pushing their native app *hard* because those users converted at double the rate of their website.
Of course they did… that’s just a self-identifying segment of the most loyal customers who valued the home-screen shortcut.
Of course they did… that’s just a self-identifying segment of the most loyal customers who valued the home-screen shortcut.
January 2, 2026 at 10:05 AM
Reminds me of a client I had pushing their native app *hard* because those users converted at double the rate of their website.
Of course they did… that’s just a self-identifying segment of the most loyal customers who valued the home-screen shortcut.
Of course they did… that’s just a self-identifying segment of the most loyal customers who valued the home-screen shortcut.
Have you compared it to building on Netlify?
If Netlify is faster, you might be better using Netlify to handle the build and just pointing Cloudflare at it.
You can use a post-build hook to purge Cloudflare.
If Netlify is faster, you might be better using Netlify to handle the build and just pointing Cloudflare at it.
You can use a post-build hook to purge Cloudflare.
January 1, 2026 at 10:12 AM
Have you compared it to building on Netlify?
If Netlify is faster, you might be better using Netlify to handle the build and just pointing Cloudflare at it.
You can use a post-build hook to purge Cloudflare.
If Netlify is faster, you might be better using Netlify to handle the build and just pointing Cloudflare at it.
You can use a post-build hook to purge Cloudflare.
Reminded me to update my PR 😉 github.com/mdn/content/...
Add note on CSP to Compression Dictionary Transport by ryantownsend · Pull Request #42487 · mdn/content
Description
Adds a note that implementing a standalone compression dictionary on websites that have a Content Security Policy requires that the connect-src directive permits the dictionary location...
github.com
December 30, 2025 at 11:36 PM
Reminded me to update my PR 😉 github.com/mdn/content/...
I’ve often wondered whether these numbers would be even tighter these days given the erosion of attention spans 🤔
December 27, 2025 at 12:04 AM
I’ve often wondered whether these numbers would be even tighter these days given the erosion of attention spans 🤔
Sending thoughts and prayers 🙏
December 22, 2025 at 10:57 AM
Sending thoughts and prayers 🙏
The only commit checks out.
December 18, 2025 at 6:50 PM
The only commit checks out.
FWIW, you can use the `zstd` binary to train an optimal dictionary automatically, see: github.com/facebook/zst...
Just be mindful of this issue: github.com/facebook/zst... – you can trim the unnecessary bytes and make it valid with `tail -c +5`
Just be mindful of this issue: github.com/facebook/zst... – you can trim the unnecessary bytes and make it valid with `tail -c +5`
facebook/zstd
Zstandard - Fast real-time compression algorithm. Contribute to facebook/zstd development by creating an account on GitHub.
github.com
December 13, 2025 at 8:31 PM
FWIW, you can use the `zstd` binary to train an optimal dictionary automatically, see: github.com/facebook/zst...
Just be mindful of this issue: github.com/facebook/zst... – you can trim the unnecessary bytes and make it valid with `tail -c +5`
Just be mindful of this issue: github.com/facebook/zst... – you can trim the unnecessary bytes and make it valid with `tail -c +5`
Nice! Are you just using one HTML document to compress another or are you using an external dictionary?
December 13, 2025 at 8:19 PM
Nice! Are you just using one HTML document to compress another or are you using an external dictionary?