TigerBeetle looks interesting. A whole networked database using statically allocated memory. I usually use that technique for small keyboard-shortcut triggered applications.
One question in case folks who work there see this:
This is the most technically impressive zig project I've seen so far. Do you have a blog post detailing your perspective on zig? i.e what design decisions of ziglang helped you in massive way, what were the warts, and in general any other thoughts too?
Yeah, I too feel that anything that's TV news should be left out of HN. I mean, as an outsider (to us politics), I didn't see a whack of difference between standard twitter/instagram slop and what was on HN, for something like say, recent US elections. It was just slop here too. Not to mention most comments just quoting tweets, making most threads "RSS for twitter".
I feel like enforcing the TV news rule better is probably the best bang for the buck.
This doesn't mean all government related stuff, for example, chatcontrol is kind of tech-relevant, so makes sense to talk about it. But really? did we need to discuss charlie kirk here, or gaza/ukraine/pahalgam/what have you? Multiple times too.
I have thought of things to do with kitty remote but have always been lazy to actually write the code. Do you have it open source by any chance for me to steal?
I hope not, my LLM detector is off if that's the case. But looking at how they've responded elsewhere it may be the case that they are pulling from LLMs to try and win arguments.
It definitely seems like they have a bunch of unconnected ways of arguing in this comment section. Like here they are arguing that the PCB design is the hardest act of engineering ever done. Elsewhere they are arguing that just seeing the schematics is all someone needs to build an iPhone.
Looking at their comment history outside this thread, I think they aren't always using LLMs. Perhaps just when something gets more technical than they have the experience for.
Yes, sqlite uses only one index per table in FROM clause... Except for when it can use the OR optimization [1]
> Left to right, no skipping, stops at the first range
I don't know if we need a soundbite for this, nor is it sqlite specific. This is a direct consequence of the fact that indexes are basically just sorted lists you can binary search over. They are sorted left to right (you have to choose some order anyhow) meaning all the other orderings cannot avail of the speedup, and thus sqlite chooses to not use it.....
Except if left column of index is measured to be low cardinality by ANALYZE [2]'s output in the sqlite_stats table, in which case it is ok to scan over it and then binary search right column. More at skip-scan optimization [3].
AI investment is more like an infrastructure investment. The telecom boom was the same, and so was the cloud boom, and to some extent, the dotcom boom too.
> Intertwined
And that is the _point_, this same pattern existed in all those previous instances too. OpenAI/Stargate and so on are just "fronts" for a team effort investment into building wildly unprofitable infrastructure costing upwards of 1T. e.g Saudi does not give a hoot about LLM technology itself, its just that there is infra being built, and once it is done, it will lock in energy demand in a massive way, and so it makes sense to help build this infra. Similarly for nvidia/gpus, which is why they invest in openai and such.
All the money people are dumping into this now, will be made back on the millions of small million dollar companies that will build products on top of this infrastructure, once it is built (which causes the money to move out of this and onto those smaller product companies - this is the market crashing/bubble bursting). I don't need to point you at the various SaaS/whatever companies making a killing on top of the extremely unprofitable cloud investments of the past, or the absolutely insane economic reach telecom has enabled today.
During this push to build the infrastructure, there will be a lot of short-term investors trying to make easy cash by greater-fool investing -- they have no intention of staying and building product companies after the crash. A portion of these that don't anticipate the crash's timing will lose lots of money.
Unfortunately, if Wall St. and such package this debt along with other debt that is bought by institutional investors/bonds/pensions, or if these guys and other wide-reaching funds buy the volatile debt themselves, then everyone who is connected to that (including grandma, including aunt's house) is exposed to that risk. This is what will decide whether it's going to be a tech bubble problem or a financial crisis.
And almost surely, the job market will dive for a while [1], since jobs depend on cash flow, and cash flow will dry up in the time period between the "infra building" state, and the "now we have high-margin products on this infra" state.
[1] The key thing is, and this is the primary problem, what "a while" is, is dependent on broader factors. The ability to pick up pace again after a bubble bursts will be dependent on the general economic health of involved countries as a whole. This is because job market diving leads to consumer/real estate/etc etc falling i.e wide reaching negative feedback loop = "macro slowdown"
One question in case folks who work there see this:
This is the most technically impressive zig project I've seen so far. Do you have a blog post detailing your perspective on zig? i.e what design decisions of ziglang helped you in massive way, what were the warts, and in general any other thoughts too?
reply