Ashley Claymore
@ashley-c.bsky.social
JavaScript Infrastructure & Tooling at Bloomberg. TC39 Delegate. London.
In my experience TA usage is low in general. More common when interacting with native/platform/wasm APIs - or using SAB with workers. i.e. at boundaries.
Ergonomics and familiarity of Arrays is strong.
Ergonomics and familiarity of Arrays is strong.
December 1, 2025 at 8:30 AM
In my experience TA usage is low in general. More common when interacting with native/platform/wasm APIs - or using SAB with workers. i.e. at boundaries.
Ergonomics and familiarity of Arrays is strong.
Ergonomics and familiarity of Arrays is strong.
I'm feeling more productive already
November 27, 2025 at 8:11 AM
I'm feeling more productive already
Thanks, Looking forward to hearing the final results!
November 25, 2025 at 10:59 AM
Thanks, Looking forward to hearing the final results!
Were there also observable reductions in creation and evaluation time?
November 25, 2025 at 10:09 AM
Were there also observable reductions in creation and evaluation time?
I did it for @jonkuperman.com but super thrilled it's having wider appeal
November 20, 2025 at 9:28 AM
I did it for @jonkuperman.com but super thrilled it's having wider appeal
All providers sharing the same app helps a little: youtu.be/JJXEmkTtuUU
How the UK is fixing its parking app nightmare
YouTube video by Chris Spargo
youtu.be
November 14, 2025 at 3:08 AM
All providers sharing the same app helps a little: youtu.be/JJXEmkTtuUU
"This miserable way
is taken by the sorry souls of those
who lived without disgrace and without praise"
is taken by the sorry souls of those
who lived without disgrace and without praise"
November 7, 2025 at 8:39 PM
"This miserable way
is taken by the sorry souls of those
who lived without disgrace and without praise"
is taken by the sorry souls of those
who lived without disgrace and without praise"
Excellent — you’re hitting the core of what a real agentic loop looks like
November 7, 2025 at 6:45 PM
Excellent — you’re hitting the core of what a real agentic loop looks like
Is this primarily for the case where the original promise was synchronously resolved?
Asking as when the promise represents actual IO adding one extra microtask seems less bad compared to the fully sync case.
Or is the hit noticeable in both cases?
Asking as when the promise represents actual IO adding one extra microtask seems less bad compared to the fully sync case.
Or is the hit noticeable in both cases?
November 6, 2025 at 7:21 AM
Is this primarily for the case where the original promise was synchronously resolved?
Asking as when the promise represents actual IO adding one extra microtask seems less bad compared to the fully sync case.
Or is the hit noticeable in both cases?
Asking as when the promise represents actual IO adding one extra microtask seems less bad compared to the fully sync case.
Or is the hit noticeable in both cases?
Interesting! Personally I appreciate (a good) template, they help me just get on with submitting the information instead of over thinking what I write.
Hadn't occurred to me that they might be a red flag for some people.
Hadn't occurred to me that they might be a red flag for some people.
November 1, 2025 at 10:41 AM
Interesting! Personally I appreciate (a good) template, they help me just get on with submitting the information instead of over thinking what I write.
Hadn't occurred to me that they might be a red flag for some people.
Hadn't occurred to me that they might be a red flag for some people.
Doctor prescribes vibe coding
October 31, 2025 at 1:40 PM
Doctor prescribes vibe coding
The song's quite catchy!
_ashley sings funk dot new_
_ashley sings funk dot new_
October 30, 2025 at 7:21 AM
The song's quite catchy!
_ashley sings funk dot new_
_ashley sings funk dot new_
Is there any clever fork de-duplication? If a state is entered into for which a fork already exists? Or maybe this "just calls out" of how derived states are already cached?
October 28, 2025 at 6:45 PM
Is there any clever fork de-duplication? If a state is entered into for which a fork already exists? Or maybe this "just calls out" of how derived states are already cached?
I also suspect many backends are not setup to do serverless scaling, and instead keep a fixed numbers of servers always up. So startup time is not a huge concern.
October 28, 2025 at 4:00 PM
I also suspect many backends are not setup to do serverless scaling, and instead keep a fixed numbers of servers always up. So startup time is not a huge concern.
Might be able to get more information by running the script via d8 and getting it to log if any deopts are occurring.
October 28, 2025 at 9:33 AM
Might be able to get more information by running the script via d8 and getting it to log if any deopts are occurring.
That is surprising. My theory would be something along the lines of the function wrapper attempts to be inlined but the console.log prevents that, so there is a de-opt cost.
Whereas the proxy is never eligible for inlining so there isn't time spent trying to optimise it.
Whereas the proxy is never eligible for inlining so there isn't time spent trying to optimise it.
October 28, 2025 at 9:33 AM
That is surprising. My theory would be something along the lines of the function wrapper attempts to be inlined but the console.log prevents that, so there is a de-opt cost.
Whereas the proxy is never eligible for inlining so there isn't time spent trying to optimise it.
Whereas the proxy is never eligible for inlining so there isn't time spent trying to optimise it.
The function wrapper is doing the extra array for loop, so doesn't seem like a fair test?
October 27, 2025 at 10:03 PM
The function wrapper is doing the extra array for loop, so doesn't seem like a fair test?
Results from mobile safari:
October 27, 2025 at 10:02 PM
Results from mobile safari: