Sam Willis
@samwillis.uk
Local First, TanStack DB, @pglite.dev and Sync Engines at @electric-sql.com.
More at https://samwillis.uk
More at https://samwillis.uk
I've got my ticket!
Who else is going to @syncconf.bsky.social?
Who else is going to @syncconf.bsky.social?
August 19, 2025 at 5:39 PM
I've got my ticket!
Who else is going to @syncconf.bsky.social?
Who else is going to @syncconf.bsky.social?
Dev tools coming soon to @tanstack.com DB!
The best client side state libs have brilliant dev tools, this is even more important for DB as you store so much state in it.
The DB dev tools will give you full visibility of all collections and live queries. 🚀🎉
The best client side state libs have brilliant dev tools, this is even more important for DB as you store so much state in it.
The DB dev tools will give you full visibility of all collections and live queries. 🚀🎉
August 15, 2025 at 1:03 PM
Dev tools coming soon to @tanstack.com DB!
The best client side state libs have brilliant dev tools, this is even more important for DB as you store so much state in it.
The DB dev tools will give you full visibility of all collections and live queries. 🚀🎉
The best client side state libs have brilliant dev tools, this is even more important for DB as you store so much state in it.
The DB dev tools will give you full visibility of all collections and live queries. 🚀🎉
Perfect lunch stop with the family after a 3hr hike! 😊
July 23, 2025 at 3:31 PM
Perfect lunch stop with the family after a 3hr hike! 😊
PGlite (WASM Postgres) now gets over 1 million downloads a week! 🤯🚀
Absolutely incredible to see how people have embraced what we have built.
Awesome work by the team!
Absolutely incredible to see how people have embraced what we have built.
Awesome work by the team!
July 14, 2025 at 3:28 PM
PGlite (WASM Postgres) now gets over 1 million downloads a week! 🤯🚀
Absolutely incredible to see how people have embraced what we have built.
Awesome work by the team!
Absolutely incredible to see how people have embraced what we have built.
Awesome work by the team!
Home for the next few weeks 😁
Love the 🏔️ in the summer!
Love the 🏔️ in the summer!
July 4, 2025 at 7:09 PM
Home for the next few weeks 😁
Love the 🏔️ in the summer!
Love the 🏔️ in the summer!
Last month I spoke @pgconf.dev about @pglite.dev, how we built it and where we plan to take it 🎉
You can now watch my talk at the link below 🚀🐘
You can now watch my talk at the link below 🚀🐘
June 9, 2025 at 2:42 PM
Last month I spoke @pgconf.dev about @pglite.dev, how we built it and where we plan to take it 🎉
You can now watch my talk at the link below 🚀🐘
You can now watch my talk at the link below 🚀🐘
Getting notified of AirPods/Tags traveling with you is smart - I like it.
But it's absurd that you can't click a button that says "these are my wife's, they will often travel with me, I don't need to know this!" - or rather "ignore forever"
But it's absurd that you can't click a button that says "these are my wife's, they will often travel with me, I don't need to know this!" - or rather "ignore forever"
May 10, 2025 at 6:50 PM
Getting notified of AirPods/Tags traveling with you is smart - I like it.
But it's absurd that you can't click a button that says "these are my wife's, they will often travel with me, I don't need to know this!" - or rather "ignore forever"
But it's absurd that you can't click a button that says "these are my wife's, they will often travel with me, I don't need to know this!" - or rather "ignore forever"
One week till my talk @pgconf.dev!
Very excited to be visiting Montreal 🇨🇦
Give me a shout if you want to meet up a chat @pglite.dev, @electric-sql.com, Sync Engines or Local-first app development.
Very excited to be visiting Montreal 🇨🇦
Give me a shout if you want to meet up a chat @pglite.dev, @electric-sql.com, Sync Engines or Local-first app development.
May 7, 2025 at 5:45 PM
One week till my talk @pgconf.dev!
Very excited to be visiting Montreal 🇨🇦
Give me a shout if you want to meet up a chat @pglite.dev, @electric-sql.com, Sync Engines or Local-first app development.
Very excited to be visiting Montreal 🇨🇦
Give me a shout if you want to meet up a chat @pglite.dev, @electric-sql.com, Sync Engines or Local-first app development.
Sunday mornings are for family reading-in-a-coffe-shop sessions 📚
May 4, 2025 at 8:23 AM
Sunday mornings are for family reading-in-a-coffe-shop sessions 📚
Got to enjoy the British summer while it's actually happening...
Moving my office to the garden for the week.
Moving my office to the garden for the week.
April 28, 2025 at 3:39 PM
Got to enjoy the British summer while it's actually happening...
Moving my office to the garden for the week.
Moving my office to the garden for the week.
PGlite is now based on Postgres 17 🎉🚀
v0.3, which we just released, is a big step forward for the project, and the foundation for a lot of exciting things coming soon!
v0.3, which we just released, is a big step forward for the project, and the foundation for a lot of exciting things coming soon!
April 24, 2025 at 3:25 PM
PGlite is now based on Postgres 17 🎉🚀
v0.3, which we just released, is a big step forward for the project, and the foundation for a lot of exciting things coming soon!
v0.3, which we just released, is a big step forward for the project, and the foundation for a lot of exciting things coming soon!
Can your Chat AI app make todo lists?
It was fun adding this feature to the demo & really shows how a sync engine ensures your UI is kept up to date with agent side effects.
PS: that token stream is multi-user via Postgres! 🤯
It was fun adding this feature to the demo & really shows how a sync engine ensures your UI is kept up to date with agent side effects.
PS: that token stream is multi-user via Postgres! 🤯
April 9, 2025 at 7:36 PM
Can your Chat AI app make todo lists?
It was fun adding this feature to the demo & really shows how a sync engine ensures your UI is kept up to date with agent side effects.
PS: that token stream is multi-user via Postgres! 🤯
It was fun adding this feature to the demo & really shows how a sync engine ensures your UI is kept up to date with agent side effects.
PS: that token stream is multi-user via Postgres! 🤯
I'm really excited to be presenting PGlite and the work it's taken to bring Postgres to WASM and the browser at
@pgconf.dev in May 🎉🐘
@pgconf.dev in May 🎉🐘
March 4, 2025 at 7:03 PM
I'm really excited to be presenting PGlite and the work it's taken to bring Postgres to WASM and the browser at
@pgconf.dev in May 🎉🐘
@pgconf.dev in May 🎉🐘
Incremental streaming joins scale proportional to the update size, not the total dataset size.
With D2TS I've found it possible to do 10s thousands updates a second with almost no change in latency as the dataset size increases. 📈🚀
With D2TS I've found it possible to do 10s thousands updates a second with almost no change in latency as the dataset size increases. 📈🚀
March 3, 2025 at 4:25 PM
Incremental streaming joins scale proportional to the update size, not the total dataset size.
With D2TS I've found it possible to do 10s thousands updates a second with almost no change in latency as the dataset size increases. 📈🚀
With D2TS I've found it possible to do 10s thousands updates a second with almost no change in latency as the dataset size increases. 📈🚀
Incremental view maintenance is seen as a black art. The tools available, such as differential dataflow, are complex, with a steep learning curve.
I want to change that with D2TS!
It has a simple pipeline api, with full type inference at every step 🚀
I want to change that with D2TS!
It has a simple pipeline api, with full type inference at every step 🚀
![const graph = new D2({ initialFrontier: 0 });
const issuesInput = graph.newInput<[string, Issue]>();
const usersInput = graph.newInput<[string, User]>();
const commentsInput = graph.newInput<[string, Comment]>();
const commentCounts = commentsInput.pipe(
map(([_key, comment]) => [comment.issue_id, 1]),
concat(issuesInput.pipe(map(([_key, issue]) => [issue.id, 0]))),
reduce((values) => {
let count = 0;
for (const [num, diff] of values) {
count += num * diff;
}
return [[count, 1]];
})
);
issuesInput.pipe(
map(([_key, issue]) => [issue.user_id, issue]),
join(usersInput.pipe(map(([_key, user]) => [user.id, user]))),
join(commentCounts),
map(([_key, [[issue, user], commentCount]]) => [
issue.id,
{
...issue,
user_id: user.id,
username: user.username,
user_email: user.email,
user_full_name: user.full_name,
comment_count: commentCount || 0,
},
]),
consolidate(),
output((msgs) => {
console.log(msgs);
})
);](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:7zsebsiw55hprbjz6prpsbem/bafkreidmctemtym2lbzvsrx47fkyaaz7gqlwwaduxobwdphffri6dbz7a4@jpeg)
March 3, 2025 at 9:28 AM
Incremental view maintenance is seen as a black art. The tools available, such as differential dataflow, are complex, with a steep learning curve.
I want to change that with D2TS!
It has a simple pipeline api, with full type inference at every step 🚀
I want to change that with D2TS!
It has a simple pipeline api, with full type inference at every step 🚀
In the above video I'm using D2TS (github.com/electric-sql...) to join three change streams coming from the Electric sync engine: Issues, Users, and Comments.
For the comments, it's calculating the count per issue.
This is all done incrementally!
For the comments, it's calculating the count per issue.
This is all done incrementally!
![const graph = new D2({ initialFrontier: 0 });
const issuesInput = graph.newInput<[string, Issue]>();
const usersInput = graph.newInput<[string, User]>();
const commentsInput = graph.newInput<[string, Comment]>();
const commentCounts = commentsInput.pipe(
map(([_key, comment]) => [comment.issue_id, 1]),
concat(issuesInput.pipe(map(([_key, issue]) => [issue.id, 0]))),
reduce((values) => {
let count = 0;
for (const [num, diff] of values) {
count += num * diff;
}
return [[count, 1]];
})
);
issuesInput.pipe(
map(([_key, issue]) => [issue.user_id, issue]),
join(usersInput.pipe(map(([_key, user]) => [user.id, user]))),
join(commentCounts),
map(([_key, [[issue, user], commentCount]]) => [
issue.id,
{
...issue,
user_id: user.id,
username: user.username,
user_email: user.email,
user_full_name: user.full_name,
comment_count: commentCount || 0,
},
]),
consolidate(),
output((msgs) => {
console.log(msgs);
})
);](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:7zsebsiw55hprbjz6prpsbem/bafkreif3eumrvvl3tsnrgrqhqlbke6co66lc3rg7ri4epz62yektej73qm@jpeg)
February 28, 2025 at 12:59 PM
In the above video I'm using D2TS (github.com/electric-sql...) to join three change streams coming from the Electric sync engine: Issues, Users, and Comments.
For the comments, it's calculating the count per issue.
This is all done incrementally!
For the comments, it's calculating the count per issue.
This is all done incrementally!
Introducing D2TS - a TypeScript implementation of Differential Dataflow I've been building 🚀
Enabling incremental computation on change streams and streaming joins, both on the server and client.
Perfect for use with the @electric-sql.com sync engine! 🧵
Enabling incremental computation on change streams and streaming joins, both on the server and client.
Perfect for use with the @electric-sql.com sync engine! 🧵
February 28, 2025 at 12:59 PM
Introducing D2TS - a TypeScript implementation of Differential Dataflow I've been building 🚀
Enabling incremental computation on change streams and streaming joins, both on the server and client.
Perfect for use with the @electric-sql.com sync engine! 🧵
Enabling incremental computation on change streams and streaming joins, both on the server and client.
Perfect for use with the @electric-sql.com sync engine! 🧵
Light weight in-memory Postgres instances with PGlite are perfect for unit tests.
We recently added a `.clone()` method to make it super easy for have a clean state for each test. Just setup the schema and fixtures once, then clone!
We recently added a `.clone()` method to make it super easy for have a clean state for each test. Just setup the schema and fixtures once, then clone!
February 24, 2025 at 12:43 PM
Light weight in-memory Postgres instances with PGlite are perfect for unit tests.
We recently added a `.clone()` method to make it super easy for have a clean state for each test. Just setup the schema and fixtures once, then clone!
We recently added a `.clone()` method to make it super easy for have a clean state for each test. Just setup the schema and fixtures once, then clone!
This video is a great dive into building a local first app & the journey @t3.gg took to choosing Dexie + MySQL.
www.youtube.com/watch?v=3gVB...
@electric-sql.com is perfect for syncing into Dexie from a Postgres backend 🐘
Being "protocol first" (http+json), you can use anything for the client store.
www.youtube.com/watch?v=3gVB...
@electric-sql.com is perfect for syncing into Dexie from a Postgres backend 🐘
Being "protocol first" (http+json), you can use anything for the client store.
February 19, 2025 at 2:50 PM
This video is a great dive into building a local first app & the journey @t3.gg took to choosing Dexie + MySQL.
www.youtube.com/watch?v=3gVB...
@electric-sql.com is perfect for syncing into Dexie from a Postgres backend 🐘
Being "protocol first" (http+json), you can use anything for the client store.
www.youtube.com/watch?v=3gVB...
@electric-sql.com is perfect for syncing into Dexie from a Postgres backend 🐘
Being "protocol first" (http+json), you can use anything for the client store.
The reality of the pride versioning scheme...
February 13, 2025 at 8:34 PM
The reality of the pride versioning scheme...
Our Postgres in WASM project, @pglite.dev, has hit 1/2 million downloads a week! 🤯
We (@electric-sql.com) have a lot planned for (native... 😉) PGlite this year, exciting times ahead!
We (@electric-sql.com) have a lot planned for (native... 😉) PGlite this year, exciting times ahead!
February 10, 2025 at 10:55 AM
Our Postgres in WASM project, @pglite.dev, has hit 1/2 million downloads a week! 🤯
We (@electric-sql.com) have a lot planned for (native... 😉) PGlite this year, exciting times ahead!
We (@electric-sql.com) have a lot planned for (native... 😉) PGlite this year, exciting times ahead!
A few years ago I build a multiplayer Sudoku app after my wife and I found ourselves enjoying racing each other over a coffee in the mornings.
Our Daughter now plays it with us 😁
Our Daughter now plays it with us 😁
February 2, 2025 at 4:09 PM
A few years ago I build a multiplayer Sudoku app after my wife and I found ourselves enjoying racing each other over a coffee in the mornings.
Our Daughter now plays it with us 😁
Our Daughter now plays it with us 😁
The top two commands in my VSCode command pallet are usually format and sort lines.
Always fun when you hit sort lines when aiming for format...
Always fun when you hit sort lines when aiming for format...
January 29, 2025 at 2:17 PM
The top two commands in my VSCode command pallet are usually format and sort lines.
Always fun when you hit sort lines when aiming for format...
Always fun when you hit sort lines when aiming for format...
My wife @keirawillis.bsky.social had an awesome 2024, she had two books published 📕📗🥳
Amazon UK have put a promo on the first in the series, for a limited time it's just 99p 🎉
If you like crime thrillers, the 80s, and ghosts... do check it out!
www.amazon.co.uk/gp/aw/d/B0CS...
Amazon UK have put a promo on the first in the series, for a limited time it's just 99p 🎉
If you like crime thrillers, the 80s, and ghosts... do check it out!
www.amazon.co.uk/gp/aw/d/B0CS...
January 1, 2025 at 5:00 PM
My wife @keirawillis.bsky.social had an awesome 2024, she had two books published 📕📗🥳
Amazon UK have put a promo on the first in the series, for a limited time it's just 99p 🎉
If you like crime thrillers, the 80s, and ghosts... do check it out!
www.amazon.co.uk/gp/aw/d/B0CS...
Amazon UK have put a promo on the first in the series, for a limited time it's just 99p 🎉
If you like crime thrillers, the 80s, and ghosts... do check it out!
www.amazon.co.uk/gp/aw/d/B0CS...