A proof can typecheck and still prove a different theorem. Coverage won't tell you — a rule too strong covers the case you started from, and more. The tell is what it now rules out. A June paper measures it: typechecking catches 41% of drifted formalisations; probing what each excludes catches 90%.
November 6, 2026 at 8:39 PM
A proof can typecheck and still prove a different theorem. Coverage won't tell you — a rule too strong covers the case you started from, and more. The tell is what it now rules out. A June paper measures it: typechecking catches 41% of drifted formalisations; probing what each excludes catches 90%.
I wrote a bit about incremental typechecking here:
dziban.net/posts/yabu-i...
in the context of the custom language, not yet of my new system.
I'd love to compare notes
dziban.net/posts/yabu-i...
in the context of the custom language, not yet of my new system.
I'd love to compare notes
Yabu Incremental Typesystem — dziban.net
Yabu is all about interactivity and the typechecking needs to be fast and furious. It also needs to deal with arbitrary staging state and we need to ensure that we merge the maximal staging subset tha...
dziban.net
November 3, 2026 at 8:25 AM
I wrote a bit about incremental typechecking here:
dziban.net/posts/yabu-i...
in the context of the custom language, not yet of my new system.
I'd love to compare notes
dziban.net/posts/yabu-i...
in the context of the custom language, not yet of my new system.
I'd love to compare notes
Would love to exchange ideas about how you've tackled some things. I'm guessing since you are using js you don't have typechecking.
One of my previous explorations of this idea involved a custom programming language for agents and now I'm trying to bring those explorations into existing langs.
One of my previous explorations of this idea involved a custom programming language for agents and now I'm trying to bring those explorations into existing langs.
Yabu Incremental Typesystem — dziban.net
Yabu is all about interactivity and the typechecking needs to be fast and furious. It also needs to deal with arbitrary staging state and we need to ensure that we merge the maximal staging subset tha...
dziban.net
November 3, 2026 at 8:25 AM
Would love to exchange ideas about how you've tackled some things. I'm guessing since you are using js you don't have typechecking.
One of my previous explorations of this idea involved a custom programming language for agents and now I'm trying to bring those explorations into existing langs.
One of my previous explorations of this idea involved a custom programming language for agents and now I'm trying to bring those explorations into existing langs.
Opus 5.5 caught a bug in my 2023 #fsharp code: prepending with :: in a recursive function, but missing List.rev. Valid types, wrong order.
Does AI’s ability to reason about intent we never formalized mark a paradigm shift?
On this bug: #typechecking 0 — #AI 1.
Does AI’s ability to reason about intent we never formalized mark a paradigm shift?
On this bug: #typechecking 0 — #AI 1.
November 1, 2026 at 2:07 PM
Opus 5.5 caught a bug in my 2023 #fsharp code: prepending with :: in a recursive function, but missing List.rev. Valid types, wrong order.
Does AI’s ability to reason about intent we never formalized mark a paradigm shift?
On this bug: #typechecking 0 — #AI 1.
Does AI’s ability to reason about intent we never formalized mark a paradigm shift?
On this bug: #typechecking 0 — #AI 1.
Also, just to explain: TMP is 90% of the reason that parsing and typechecking is as complex as it is.
October 25, 2026 at 3:20 PM
Also, just to explain: TMP is 90% of the reason that parsing and typechecking is as complex as it is.
me learning lisp: whoa parsing isn’t separate from execution wilddddd how cool
me learning c++: whoa parsing isn’t separate from typechecking oh fuck oh fuck oh fuck no no no not like this do not mix the stages like this DO NOT—
*the c++ standards committee has had me assassinated mid rant*
me learning c++: whoa parsing isn’t separate from typechecking oh fuck oh fuck oh fuck no no no not like this do not mix the stages like this DO NOT—
*the c++ standards committee has had me assassinated mid rant*
eeee okieeeee
first horror i just recalled
C++ does not afaict and iirc have a meaningfully separate parsing and typechecking phase
i tried to find one in the spec i looked i looked i looked
but afaict they are LITERALLY INTERTWINGLED
this hurt me spiritually for a good while ada
first horror i just recalled
C++ does not afaict and iirc have a meaningfully separate parsing and typechecking phase
i tried to find one in the spec i looked i looked i looked
but afaict they are LITERALLY INTERTWINGLED
this hurt me spiritually for a good while ada
October 25, 2026 at 3:18 PM
me learning lisp: whoa parsing isn’t separate from execution wilddddd how cool
me learning c++: whoa parsing isn’t separate from typechecking oh fuck oh fuck oh fuck no no no not like this do not mix the stages like this DO NOT—
*the c++ standards committee has had me assassinated mid rant*
me learning c++: whoa parsing isn’t separate from typechecking oh fuck oh fuck oh fuck no no no not like this do not mix the stages like this DO NOT—
*the c++ standards committee has had me assassinated mid rant*
eeee okieeeee
first horror i just recalled
C++ does not afaict and iirc have a meaningfully separate parsing and typechecking phase
i tried to find one in the spec i looked i looked i looked
but afaict they are LITERALLY INTERTWINGLED
this hurt me spiritually for a good while ada
first horror i just recalled
C++ does not afaict and iirc have a meaningfully separate parsing and typechecking phase
i tried to find one in the spec i looked i looked i looked
but afaict they are LITERALLY INTERTWINGLED
this hurt me spiritually for a good while ada
October 25, 2026 at 3:15 PM
eeee okieeeee
first horror i just recalled
C++ does not afaict and iirc have a meaningfully separate parsing and typechecking phase
i tried to find one in the spec i looked i looked i looked
but afaict they are LITERALLY INTERTWINGLED
this hurt me spiritually for a good while ada
first horror i just recalled
C++ does not afaict and iirc have a meaningfully separate parsing and typechecking phase
i tried to find one in the spec i looked i looked i looked
but afaict they are LITERALLY INTERTWINGLED
this hurt me spiritually for a good while ada
Relevant XKCD.
s/compiling/typechecking/
s/compiling/typechecking/
October 23, 2026 at 11:32 AM
Relevant XKCD.
s/compiling/typechecking/
s/compiling/typechecking/
Bun RAD Studio v1.5.2 now runs on Bun v1.4.4! 🚀
• Native bun fmt in-editor formatting (+ TOML, Svelte, MDX, HBS, MJML)
• Fast bun check typechecking
• Bun.sql query cancel & 0-row column metadata
• bun:sqlite UDFs, aggregates & update hooks
github.com/codecaine-zz...
#typescript #files #games #bun
• Native bun fmt in-editor formatting (+ TOML, Svelte, MDX, HBS, MJML)
• Fast bun check typechecking
• Bun.sql query cancel & 0-row column metadata
• bun:sqlite UDFs, aggregates & update hooks
github.com/codecaine-zz...
#typescript #files #games #bun
GitHub - codecaine-zz/bun_rad_studio: A high-performance Rapid Application Development (RAD) Visual IDE for Bun and Webview-Bun, inspired by classic Borland Delphi and Visual Basic 6, built with moder...
A high-performance Rapid Application Development (RAD) Visual IDE for Bun and Webview-Bun, inspired by classic Borland Delphi and Visual Basic 6, built with modern web technologies. - codecaine-zz/...
github.com
October 27, 2026 at 1:44 PM
Bun RAD Studio v1.5.2 now runs on Bun v1.4.4! 🚀
• Native bun fmt in-editor formatting (+ TOML, Svelte, MDX, HBS, MJML)
• Fast bun check typechecking
• Bun.sql query cancel & 0-row column metadata
• bun:sqlite UDFs, aggregates & update hooks
github.com/codecaine-zz...
#typescript #files #games #bun
• Native bun fmt in-editor formatting (+ TOML, Svelte, MDX, HBS, MJML)
• Fast bun check typechecking
• Bun.sql query cancel & 0-row column metadata
• bun:sqlite UDFs, aggregates & update hooks
github.com/codecaine-zz...
#typescript #files #games #bun
compaction for idle Claude threads before prompt cache expires
t3 thread CLI to start, read, and message threads from the terminal
- PR diff badges that split line counts into code, tests, and docs
- migrating T3 Code's own typechecking onto tsc-rs
- purged over 50% of our test suite
- (and roughly
t3 thread CLI to start, read, and message threads from the terminal
- PR diff badges that split line counts into code, tests, and docs
- migrating T3 Code's own typechecking onto tsc-rs
- purged over 50% of our test suite
- (and roughly
October 25, 2026 at 3:27 AM
compaction for idle Claude threads before prompt cache expires
t3 thread CLI to start, read, and message threads from the terminal
- PR diff badges that split line counts into code, tests, and docs
- migrating T3 Code's own typechecking onto tsc-rs
- purged over 50% of our test suite
- (and roughly
t3 thread CLI to start, read, and message threads from the terminal
- PR diff badges that split line counts into code, tests, and docs
- migrating T3 Code's own typechecking onto tsc-rs
- purged over 50% of our test suite
- (and roughly
and docs
- migrating T3 Code's own typechecking onto tsc-rs
- purged over 50% of our test suite
- (and roughly 370 other merged PRs)
Pretty slow month. Hope I can pick up the pace for the rest of Q4!
- migrating T3 Code's own typechecking onto tsc-rs
- purged over 50% of our test suite
- (and roughly 370 other merged PRs)
Pretty slow month. Hope I can pick up the pace for the rest of Q4!
October 25, 2026 at 3:29 AM
and docs
- migrating T3 Code's own typechecking onto tsc-rs
- purged over 50% of our test suite
- (and roughly 370 other merged PRs)
Pretty slow month. Hope I can pick up the pace for the rest of Q4!
- migrating T3 Code's own typechecking onto tsc-rs
- purged over 50% of our test suite
- (and roughly 370 other merged PRs)
Pretty slow month. Hope I can pick up the pace for the rest of Q4!
https://github.com/ChrisVine/js_of_ocaml-and-gjs
seems like someone has done this already!
i wouldn't have any typechecking but oh well
seems like someone has done this already!
i wouldn't have any typechecking but oh well
GitHub - ChrisVine/js_of_ocaml-and-gjs
Contribute to ChrisVine/js_of_ocaml-and-gjs development by creating an account on GitHub.
github.com
October 25, 2026 at 12:34 AM
https://github.com/ChrisVine/js_of_ocaml-and-gjs
seems like someone has done this already!
i wouldn't have any typechecking but oh well
seems like someone has done this already!
i wouldn't have any typechecking but oh well
… Fai programs can compile, provide tooling-based information, and run in milliseconds, and we make sure their compilation time and typechecking time don't grow based on the size of the project but only based on the size of the edit done since the last compilation...
October 16, 2026 at 2:50 PM
… Fai programs can compile, provide tooling-based information, and run in milliseconds, and we make sure their compilation time and typechecking time don't grow based on the size of the project but only based on the size of the edit done since the last compilation...
i like my lua with typechecking
October 15, 2026 at 1:02 AM
i like my lua with typechecking
Mingkun Xiao, Yixuan Sun: Typechecking without Minimum Types: Finite Bases and Backward Checking for $F_{<:}^{\top}$ https://arxiv.org/abs/2610.20356 https://arxiv.org/pdf/2610.20356 https://arxiv.org/html/2610.20356
October 15, 2026 at 6:43 AM
Mingkun Xiao, Yixuan Sun: Typechecking without Minimum Types: Finite Bases and Backward Checking for $F_{<:}^{\top}$ https://arxiv.org/abs/2610.20356 https://arxiv.org/pdf/2610.20356 https://arxiv.org/html/2610.20356
Thank you Opus 5.5 🫡
Kinda wild that I vibe coded a slop port of the LANGUAGE ITSELF and got a 3x reduction in typechecking
Kinda wild that I vibe coded a slop port of the LANGUAGE ITSELF and got a 3x reduction in typechecking
October 12, 2026 at 11:05 AM
Thank you Opus 5.5 🫡
Kinda wild that I vibe coded a slop port of the LANGUAGE ITSELF and got a 3x reduction in typechecking
Kinda wild that I vibe coded a slop port of the LANGUAGE ITSELF and got a 3x reduction in typechecking
Thank you Opus 5.5 🫡
Kinda wild that I vibe coded a slop port of the LANGUAGE ITSELF and got a 3x reduction in typechecking
Kinda wild that I vibe coded a slop port of the LANGUAGE ITSELF and got a 3x reduction in typechecking
October 12, 2026 at 10:45 AM
Thank you Opus 5.5 🫡
Kinda wild that I vibe coded a slop port of the LANGUAGE ITSELF and got a 3x reduction in typechecking
Kinda wild that I vibe coded a slop port of the LANGUAGE ITSELF and got a 3x reduction in typechecking
it cannot be healthy to be so excited so continuously
like if it weren’t for the fact that this shit is typechecking, I would be worried about mania
but it’s checking, and it is grounded in citations
and it all makes good obvious math sense
in any case this wave will end soon somehow or other!
like if it weren’t for the fact that this shit is typechecking, I would be worried about mania
but it’s checking, and it is grounded in citations
and it all makes good obvious math sense
in any case this wave will end soon somehow or other!
October 9, 2026 at 5:44 PM
it cannot be healthy to be so excited so continuously
like if it weren’t for the fact that this shit is typechecking, I would be worried about mania
but it’s checking, and it is grounded in citations
and it all makes good obvious math sense
in any case this wave will end soon somehow or other!
like if it weren’t for the fact that this shit is typechecking, I would be worried about mania
but it’s checking, and it is grounded in citations
and it all makes good obvious math sense
in any case this wave will end soon somehow or other!
Today I'm trying to understand why some people have gone to extreme lengths to avoid function colored language features
Having lived with Kotlin coroutines for a while now, I have more often wished that stack-affine typechecking (a term I made up just now) were more generic than
Having lived with Kotlin coroutines for a while now, I have more often wished that stack-affine typechecking (a term I made up just now) were more generic than
October 5, 2026 at 8:28 PM
Today I'm trying to understand why some people have gone to extreme lengths to avoid function colored language features
Having lived with Kotlin coroutines for a while now, I have more often wished that stack-affine typechecking (a term I made up just now) were more generic than
Having lived with Kotlin coroutines for a while now, I have more often wished that stack-affine typechecking (a term I made up just now) were more generic than
Most software isn't formally proven correct, but there are lots of techniques that have been developed for building confidence and increasingly strong guarantees about the correctness of a system. Testing, typechecking, proving a subset of the program, etc.
October 3, 2026 at 6:24 PM
Most software isn't formally proven correct, but there are lots of techniques that have been developed for building confidence and increasingly strong guarantees about the correctness of a system. Testing, typechecking, proving a subset of the program, etc.
"Contributing upstream: Both projects are active and easy to contribute to. While working on the migration I hit a bug in Pyrefly, and I recently fixed it upstream. It was a nice feeling to give something back to the tool we now depend on." ❤️🔥🥲
napari.org/island-dispa...
napari.org/island-dispa...
From Any to Certainty: A Typechecking Journey — napari blog documentation
napari.org
September 24, 2026 at 1:38 AM
"Contributing upstream: Both projects are active and easy to contribute to. While working on the migration I hit a bug in Pyrefly, and I recently fixed it upstream. It was a nice feeling to give something back to the tool we now depend on." ❤️🔥🥲
napari.org/island-dispa...
napari.org/island-dispa...
Thanks for the shout out! 😍 Our goal in writing this blog was to share the (recent) overthinking we've done our typechecking journey to help others. Thank you Aniket for writing this up! Our new triage team truly shined 🌠 on this multi-week process: napari.org/stable/devel...
September 23, 2026 at 8:45 PM
Thanks for the shout out! 😍 Our goal in writing this blog was to share the (recent) overthinking we've done our typechecking journey to help others. Thank you Aniket for writing this up! Our new triage team truly shined 🌠 on this multi-week process: napari.org/stable/devel...
@napari.org reports its Pyrefly check at ~0.7s warm / 0.9s cold, versus ~15–25s cold in its previous mypy setup.
Fast checks change habits: run it locally, fix the thing, move on. Compute left for science.
Thanks for the great write-up on your experience!
napari.org/island-dispa...
Fast checks change habits: run it locally, fix the thing, move on. Compute left for science.
Thanks for the great write-up on your experience!
napari.org/island-dispa...
From Any to Certainty: A Typechecking Journey — napari blog documentation
napari.org
September 23, 2026 at 4:54 PM
@napari.org reports its Pyrefly check at ~0.7s warm / 0.9s cold, versus ~15–25s cold in its previous mypy setup.
Fast checks change habits: run it locally, fix the thing, move on. Compute left for science.
Thanks for the great write-up on your experience!
napari.org/island-dispa...
Fast checks change habits: run it locally, fix the thing, move on. Compute left for science.
Thanks for the great write-up on your experience!
napari.org/island-dispa...
ioctl() is basically slipping a tenner to the bouncer to let an unchecked bag of bytes in or out (or both!) of kernel space. It's the void* of typechecking/security.
Paint it /proc and make every driver implement their own hand-rolled sscanf() and sprintf() ASCII string parsers? Sure. Why not.
Paint it /proc and make every driver implement their own hand-rolled sscanf() and sprintf() ASCII string parsers? Sure. Why not.
September 21, 2026 at 5:56 AM
ioctl() is basically slipping a tenner to the bouncer to let an unchecked bag of bytes in or out (or both!) of kernel space. It's the void* of typechecking/security.
Paint it /proc and make every driver implement their own hand-rolled sscanf() and sprintf() ASCII string parsers? Sure. Why not.
Paint it /proc and make every driver implement their own hand-rolled sscanf() and sprintf() ASCII string parsers? Sure. Why not.
the typechecking happens during continuous integration
September 18, 2026 at 5:16 PM
the typechecking happens during continuous integration