ccleve

It's not an index on every column. It's a single index across all columns.

Sorry, no. The original schema was correct, and the new one is a mistake.

The reason is that the new schema adds a great deal of needless complexity, requires the overhead of foreign keys, and makes it a hassle to change things later.

It's better to stick the the original design and add a unique index with key prefix compression, which all major databases do these days. This means that the leading values gets compressed out and the resulting index will be no larger and no slower than the one with foreign keys.

If you include all of the keys in the index, then it will be a covering index and all queries will hit the index only, and not the heap table.

Postgres does something similar. Every time a query returns a row the system must do a lookup into a visibility map to see if the row is visible to the current transaction. It can also check if it's a completely dead row. If there is an index involved, then on the next row fetch Postgres will report to the index whether the last row was dead or not. The index can elect to delete the row reference, which means a write during a select statement.

Enterprise search is a hard problem, but this looks pretty slick.

What's your tech stack?

Did you create the integrations from scratch, or use something like Zapier?

Good idea. Bubble appears to have multiple such templates.

It appears that all these extra functions are written in Rust, but the core Timescale product is written in C. Why did you do it that way? Would you do it differently if you were starting from scratch today?

It does make sense. Thanks.

Duplicating my question to Lockerman below. Sorry for the double post.

Is the -> operator just syntactic sugar? Is this:

SELECT device_id, timevector(ts, val) -> sort() -> delta() -> abs() -> sum() as volatility

the same as this?

SELECT device_id, sum(abs(delta(sort(timevector(ts, val))))) as volatility

Is this just syntactic sugar, then?

Is this:

SELECT device_id, timevector(ts, val) -> sort() -> delta() -> abs() -> sum() as volatility

the same as this?

SELECT device_id, sum(abs(delta(sort(timevector(ts, val))))) as volatility

The big thing to notice here: all of the top themes are minimalist. Limited colors, simple layout, just nice spacing and typography.

I've spent a lot of time looking for good, professional-looking minimalist themes and the pickings have been slim, regardless of platform. Too many theme designers load up their designs with unnecessary elements.

Note to designers: less is more.

"Illegal" means contrary to the law. Plenty of things that are illegal but carry no criminal penalties.

As a practical matter, the court would rule that sham candidates are impermissible, and then their opponents would have an opportunity to remove them from the ballot via a civil lawsuit, or seek sanctions again whoever put them up, or collect civil damages, or in an extreme case overturn an election altogether.

A friend of mine (and my successor as chairman of the Chicago Republican Party) filed a case in the U.S. Supreme Court this week over the same tactic right here in Chicago.

https://www.chicagotribune.com/politics/ct-michael-madigan-c...

If we win this thing, then filing sham candidates will become illegal in the U.S. It's an utterly corrupt practice.

Greg, if you see this, what is your opinion of the newer you-can-use-this-software-for-anything-but-cloud-services licenses?

https://www.elastic.co/pricing/faq/licensing

https://www.mongodb.com/licensing/server-side-public-license...

If you were building a product from scratch today, what business model would you choose? (Assume we're building horizontal infrastructure software, like Pinecone, and we'll offer a SaaS option.) a) Open source SSPL, b) Closed source commercial license for offsite use, c) Just straight SaaS.

I wish I could hire non-technicals. The problem is that the product is only partially built. No product means no need for anything other than product at this point.

No one has mentioned it, although it may well be a factor.

I think it's not so much the politics as it is that people see other interests and don't know what to do with them. It's as if having done additional things during your life somehow dilutes the main things.

Don't even get to the interview stage. My tech skills never got assessed.

See my other comment. Yes, we will hire actual old people.

We'll hire more than one. A bunch, if this product works out.

I wrote the ad. Holy cow, I didn't expect to be #1 on Hacker News. Will answer any questions.

Here's one you might ask: Why? I'm pushing 60 and I've attempted to interview for developer jobs over the last year. Got nowhere despite 40 years of experience. Getting really, really tired of this industry's attitude toward people like me.

Based on questions below, I'll give a bit more. Earlier this year I had a personal referral into Facebook for a developer position. They do hiring centrally. A very young person in the HR department spoke to me, put me in the central database, and apparently shopped my experience around to a number of different groups. No takers. No one wanted to interview me, so the application died. Never did a tech assessment.

So now I've started a new group, and I expect this thing is going to generate a lot more money than any work at Facebook ever would. So Facebook did me a favor. Still, it rankles.

Crunchbase says 101-250 employees, so they must have guessed wrong. https://www.crunchbase.com/organization/turbot Hope you get to that level soon, looks like a great product.

Oh, I see. The parent company, Turbot, has 100+ employees and has been at this for a while.

Your website says 200+ data sources. How the heck did you generate so many? This project does not look very old.

Yes, true. If your comparisons involve pointer-chasing or string comparison or anything complex, then it's not clear which strategy is faster. I always write the simplest possible code first, which is usually a linear search, benchmark it, and then try to beat the benchmark if it's warranted.

It's a cool idea, but you have to keep in mind that binary searches are not optimal for b-tree nodes unless the node is really big. If you have a smaller node, < 32 or 64 elements, a simple linear search is faster. The reason has something to do with processor cache, branch mis-prediction, and pipelining.

A linear search can also be optimized to use SIMD, which is a big win.

If you really want to use a path hint, you could just say "start my linear search at the point where I left off last time", and if the first element is too big, wrap around to the beginning. But that might not buy you much.

What's your point? That it's an "interesting take"?

Or that you just wanted to level a personal attack on me?

Um, what? I am talking about what happened. It is others who are motivated to criticize Pai not because of what happened in this specific instance, but because of general disagreement with him over other issues (like net neutrality).

This is an absurd hit piece. I promise you that Ajit Pai did not personally review each census block to determine if it needed service or not. That happens at the staff level.

Plus, the assertion in the headline that Pai "mismanaged $9 billion fund" is bloviation -- the FCC is simply asking some grant recipients to return a small portion of the money (6%) based on revised maps. Honestly, do we need to politicize everything?

Complete, utter garbage. Cites no statistics.

The truth is, about 25% of new businesses in the US are started by first-generation immigrants. 30% of new business owners only have a high-school education. Here are a few more stats: https://www.smallbizgenius.net/by-the-numbers/entrepreneur-s...

Plus, it's way better to generate a UUID in the app server and send it to the server as a string. The reason is that you can deal with the id as a plain string and don't have to deal with a non-standard datatype in your app.

The Postgres JDBC driver does not guarantee that batch inserts come back in the same order that you insert them (when you use RETURNING *). So, if you generate UUIDs server-side, you can't conveniently match them up with the records you just inserted. You're better off generating them in the app server first and then sending them to Postgres.