I guess the best way to render it would be exactly like `` when closed, but when opened there should be a search box where it'd normally show the first option.

Clicking search would call a javascript function, which could then adjust the set of available options via the standard DOM APIs.

Alternatively they could add support for a special element `` which can contain whatever you want, which would render fixed at the when the drop down is opened and can be interacted with like normal html elements. Then you could place your own search fields inside.

]]>Thu, 20 Aug 2026 17:20:07 +0000https://news.ycombinator.com/item?id=49377496CodesInChaoshttps://news.ycombinator.com/item?id=49377496https://news.ycombinator.com/item?id=49377496What I'd like to see are searchable drop-downs (not to be confused with text input with suggestions).

And the date/time picker controls are too limited for many applications.

]]>Thu, 20 Aug 2026 16:28:56 +0000https://news.ycombinator.com/item?id=49376858CodesInChaoshttps://news.ycombinator.com/item?id=49376858https://news.ycombinator.com/item?id=49376858Unlogged tables still support MVCC and keep old versions of tuples around until they're garbage collected.

]]>Thu, 20 Aug 2026 15:51:42 +0000https://news.ycombinator.com/item?id=49376340CodesInChaoshttps://news.ycombinator.com/item?id=49376340https://news.ycombinator.com/item?id=49376340In our multi-tenant e-commerce application we handled it like this:

* Metadata for each file is stored in the database

* The application accesses files through an abstraction based on that metadata, and doesn't care if the actual file is stored in S3 or the database

* File types which are small and few, are stored in the database. For example letterheads, logos, terms-and-conditions. (a few gigabytes total)

* File types which are big (e.g. CSV-reports) or many (e.g. invoice-PDFs) are stored in S3 (several terabytes total)

* Development and test systems often use the database for everything and don't have an associated S3 bucket.

* Most of the application remains usable without S3 (access to invoice PDFs and CSV-reports is not critical)

In our case the DB was Mongo, but I expect Postgres to work the same.

We actually stored all files in the database originally, and only migrated after it reached several terabytes. It worked perfectly fine, but was rather expensive. So next time I'd go for S3 for the start.

]]>Thu, 20 Aug 2026 15:35:56 +0000https://news.ycombinator.com/item?id=49376131CodesInChaoshttps://news.ycombinator.com/item?id=49376131https://news.ycombinator.com/item?id=49376131My rule is "use a single database for as many of you persistence needs as possible". Often Postgres is a good choice for that database, but many other general purpose DBMSs will work just as well.

In my previous company we used MongoDB, for almost everything, including text logs, request logs, job-queues and small template files. We added S3 since storing terabytes of files in a database is expensive. Now, I wouldn't choose MongoDB for a new app, since the data model and query language suck, but it performed reasonably across many use-cases.

For my next application I will use Postgres as primary database. Probably will integrate S3 before going live, to avoid the necessary data migration later, but haven't decided yet if that's a premature optimization.

]]>Thu, 20 Aug 2026 15:06:18 +0000https://news.ycombinator.com/item?id=49375707CodesInChaoshttps://news.ycombinator.com/item?id=49375707https://news.ycombinator.com/item?id=49375707One specialist and a group of generalists is often enough. Especially if you are allowed to contact the specialist outside normal working hours in rare emergencies.

]]>Thu, 20 Aug 2026 14:48:55 +0000https://news.ycombinator.com/item?id=49375411CodesInChaoshttps://news.ycombinator.com/item?id=49375411https://news.ycombinator.com/item?id=49375411I wouldn't put auch limitations in the system prompt. A mix of fine-tuning and out-of-band detection appears to be a better fit.

]]>Sun, 16 Aug 2026 15:46:37 +0000https://news.ycombinator.com/item?id=49321116CodesInChaoshttps://news.ycombinator.com/item?id=49321116https://news.ycombinator.com/item?id=49321116> apparently the Japanese did not like Unicode's tendencies towards Aristotelian essentialism

Do you mean Han Unification? I.e. that conceptually equivalent characters which are written differently in Japan and China received only a single unicode code-point, and are rendered the Chinese way by default on most computers?

]]>Sat, 15 Aug 2026 17:34:20 +0000https://news.ycombinator.com/item?id=49312502CodesInChaoshttps://news.ycombinator.com/item?id=49312502https://news.ycombinator.com/item?id=49312502> just use SSE

And then the user opens your website in a handful of tabs, and everything breaks because having enough open SSE connections blocks ordinary http requests to that origin.

You can avoid that by using a shared worker for all your tabs, but then you lose the simplicity advantage.

]]>Thu, 13 Aug 2026 07:53:18 +0000https://news.ycombinator.com/item?id=49282926CodesInChaoshttps://news.ycombinator.com/item?id=49282926https://news.ycombinator.com/item?id=49282926It's amusing how top law students become glorified narrators/voice actors.

Had to visit one of these every time we raised capital. He never did more than read the text aloud and perhaps correct a typo. I don't want to know how much the company paid for that privilege.

]]>Wed, 12 Aug 2026 15:19:56 +0000https://news.ycombinator.com/item?id=49273803CodesInChaoshttps://news.ycombinator.com/item?id=49273803https://news.ycombinator.com/item?id=49273803They have google.com, which is the brand they use towards customers. The Alphabet brand is little more than a stock ticker.

]]>Wed, 12 Aug 2026 15:16:18 +0000https://news.ycombinator.com/item?id=49273731CodesInChaoshttps://news.ycombinator.com/item?id=49273731https://news.ycombinator.com/item?id=49273731Which gTLDs are reliable besides the classics (com, net, org,...)? By that I mean that the TLD owner won't screw you over a few years from now, for example by raising prices?

I kind of like the idea to use .com for the marketing site, and .app for the actual web-app.

]]>Wed, 12 Aug 2026 15:13:08 +0000https://news.ycombinator.com/item?id=49273675CodesInChaoshttps://news.ycombinator.com/item?id=49273675https://news.ycombinator.com/item?id=49273675I understand using http on localhost and in some cases local domains accessible within the LAN. But I don't see much point in using http for public domains, especially since browsers apply many restrictions to insecure origins (http://*.localhost is considered secure).

]]>Wed, 12 Aug 2026 15:09:23 +0000https://news.ycombinator.com/item?id=49273616CodesInChaoshttps://news.ycombinator.com/item?id=49273616https://news.ycombinator.com/item?id=49273616A lot of things go up forever, as long as you denominate them in an inflationary currency ;-)

]]>Tue, 11 Aug 2026 15:29:15 +0000https://news.ycombinator.com/item?id=49259883CodesInChaoshttps://news.ycombinator.com/item?id=49259883https://news.ycombinator.com/item?id=49259883Yes, accounting/tax related stuff is still annoying. But there are a few more advantages:

* Employee share options are taxed when they sell the shares, not when issuing the option (regardless of strike price) nor when exercising it.

* Changes to the shareholder structure (e.g. raising capital) don't require visiting a notary

And I'm not sure if incorporating in a different EU country (while still looking just like a German EU Inc) would avoid some of the German pain points.

]]>Tue, 11 Aug 2026 08:55:08 +0000https://news.ycombinator.com/item?id=49255133CodesInChaoshttps://news.ycombinator.com/item?id=49255133https://news.ycombinator.com/item?id=49255133I really hope "EU Inc." will reduce the pain in that area.

]]>Tue, 11 Aug 2026 08:39:46 +0000https://news.ycombinator.com/item?id=49255030CodesInChaoshttps://news.ycombinator.com/item?id=49255030https://news.ycombinator.com/item?id=49255030They can't be as bad as KfW, right?

]]>Tue, 11 Aug 2026 08:22:28 +0000https://news.ycombinator.com/item?id=49254925CodesInChaoshttps://news.ycombinator.com/item?id=49254925https://news.ycombinator.com/item?id=492549251. You can move those folders to other locations. I used to do that in the Windows XP days.

2. They are required because applications can ask Windows "what is the 'music' folder" and get back the path. Most of those applications won't expect a missing folder.

]]>Tue, 11 Aug 2026 06:33:09 +0000https://news.ycombinator.com/item?id=49254140CodesInChaoshttps://news.ycombinator.com/item?id=49254140https://news.ycombinator.com/item?id=49254140The patent office applies a much lower "non-obviousness" threshold than how a technical person would interpret that word.

]]>Mon, 10 Aug 2026 17:23:13 +0000https://news.ycombinator.com/item?id=49246823CodesInChaoshttps://news.ycombinator.com/item?id=49246823https://news.ycombinator.com/item?id=49246823