Hacker News Reader

Comment by azemetre | Hacker News Reader

Does SQLite talk about how they plan to exist beyond 2050 across multiple lifetimes?

Not trying to be chide but it seems like with such a young industry we need better social tools to make sure this effort is preserved for future devs.

Churn has been endemic in our industry and it has probably held us back a good 20 years.

View on HN

As the time horizon increases, planning for the future is necessary, then prudent, then sensible, then optimistic, then aspirational, then foolish, then sheer arrogance. Claiming 25 years of support for something like SQLite is already on the farther end of the good set of those adjectives as it is. And I don't mean that as disrespect for that project; that's actually a statement of respect because for the vast majority of projects out there I'd put 25 years of support as already being at "sheer arrogance", so putting them down somewhere around "optimistic" is already high praise. Claiming they've got a 50 or 100 year plan might sound good but it wouldn't mean anything real.

What they can do is renew the promise going forward; if in 2030 they again commit to 25 years of support, that would mean something to me. Claiming they can promise to be supporting it in 2075 or something right now is just not a sensible thing to do.

Having a plan for several hundred years is possible and we've seen such things happen in other facets of life. We as humans are clearly capable of building robust durable social organizations, religion and civics both being testaments.

I'm curious how these plans would look and work in the context of software development. That was more what my question is about (also only being familiar with sqlite taking this seriously).

We've seen what lawyers can accomplish with their BAR associations and those were created over 200 years ago in the US! Lawyers also work with one of the clunkiest DSLs ever (legalese).

Imagine what they could accomplished if they used an actual language. :D

I’d be interested to know what you would classify as having been planned to last hundreds of years. Most of the long term institutions I can think of are the results of inertia and evolution, having been set up initially as an expediency in their time, rather than conforming to a plan set out hundreds of years ago.

The Philadelphia BAR Association was established in ~1800. I doubt the profession of law is going to disappear anytime soon, and lawyers have done a good job building their profession all things considered. Imagine if the only way you could legally sell software was through partnerships with other developers?

Do you think such a thing would have helped or hurt our industry?

I honestly think help.

What I mean is that the bar was set up for the lawyers themselves at that time. They didn’t create a 250 year plan for a Philadelphia bar that has played out in all that time and gotten us to today. It’s stayed in existence because it happened to stay useful for the lawyers that followed after them. Law itself is a collection of decisions made by judges and juries in trials, not decisions that are calibrated to have an impact over hundreds of years. Institutions are more like organisms that evolve, trying to adapt to the environment they find themselves in. The ones that work are able to stick around, and the ones that don’t die off.

You don't see an institution that established useful norms persisting for lifetimes as one worth preserving and emulating?

I do.

Medieval guilds are another equivalent but they could not deal with the industrial revolution or colonialism, so they don't seem like something worth studying (outside of their failures) if it can't deal with societal change.

You’re missing what I’m saying. I’m not commenting at all on their usefulness. I’m saying that the motivating factors that build and grow institutions are short term. Institutions last because people happen to find them useful over successive short time horizons, or they’re able to change them to suit the needs of the time. There’s no super long term planning in it, some happened to have the right combination of elements and some didn’t.

They should write a book on "Design and Implementation of SQLite". And make a course as well. That would interest a lot of people and ensure future generations pick up where they decided to retire.

I do think this is a good approach for many open source projects.

Always thought neovim should do something like this. How to recreate the base of neovim, or how to recreate popular plugins with minimal lua code.

Got to wonder how more sustainable that would be versus relying on donations.

I’m trying and failing to think of another free software product I honestly expect to still work on my current data past 2050. And this isn’t good enough?

It's good, but this also assume that the people taking care of this product in the future (which may not even born right now) will hold the same attitudes.

How do we plan to make sure the lessons we've learned during development now will still be taught 300 years from now?

I'm not putting the onus on sqlite to solve this but they are also the only organization I know of that is taking the idea seriously.

Just more thinking in the open and seeing what other people are trying to solve similar problems (ensure teachings continue past their lives) outside the context of universities.

Thinking like this is how we ended up with a panic about Y2K. Programmers in the 1970s and 80s could not conceive that their code would still be running in 2000.

The computing industry was in a huge amount of flux in the 1970s. How many bits are in a byte? Not a settled question. Code was being rewritten for new platforms all the time. Imagining that some design decisions would last for decades probably seemed laughable.

Some churn is fads, but some is legitimate (e.g. "we know how to do this better now".) Every living system is bound to churn, and that's a good thing, because it means we're learning how to do things better. I'm happy to have rust and typescript, for instance, even though they represent some amount of churn for c and javascript.

The easiest way would be to just write a spec for the data format, which I think they already have?

If any tooling fails in 25 years, you can at least write a new program to get your data back out. Then you can import it into the next hot thing.

Reply on news.ycombinator.com