Ask HN: Would you choose Linux or BSD if you were building a new laptop for devs

No, don't develop your own OS. Then you'd get your own weird-ass badly maintained OS instead of the standard Ubuntu or Red Hat or what have you. Instead just offer two or three distros preinstalled, and offer (maintained!) installation guides for when people want to do a fresh reinstall. Any weird drivers should be easily installable with a single-line yum or apt-get.

This is exactly what some cloud providers do, like DigitalOcean or Linode. They don't develop their own OS, they offer the stuff that people love and expect.

Also, I'd expect the number of BSD users to be a single percentage of the Linux users. I can't see that being a viable business, but it's worth some research.

If it were 25 years ago I'd say 'Definitely BSD'.

Today, it's 'Definitely Linux'.

"Don't reinvent" the wheel." - "If it works, don't fix it" - "Never teach your grandmother to suck eggs" - etc - etc - etc: Never fall victim to the NIH (Not Invented Here) Syndrome. Bill gates had it, it ruined Microsoft. In other words, don't think you can build a better OS than hundreds/thousands of other brains can.

> Now we need to create a new operating system for an optimal user experience.

No we do not.

TLDR: Choose Pop!_OS or ElementaryOS as they both match the 'It just works' requirement for all users.

Since you are targeting developers, it already indicates to me that you build on top of Linux which meets the drivers requirement to support the laptop's hardware. Linux already has the majority of the developer mind-share, so I see no reason to use a BSD for the desktop in this particular use-case. Don't even think about rolling your own distro unless you have a good reason to or you can afford to maintain it.

Now, the real question is: Which distro to choose for developers for optimal user experience?

To help you choose such a distro, there has to be one that must offer a level of customization that doesn't sacrifice the consistency of the OS with standard defaults to revert back to via both a CLI or GUI settings control (As found in macOS and Windows). Another requirement is having a distro that has software that feels integrated with the system software and it behaves predictably such as desktop keybinding software must work across all windows and apps or the availability of accessibility software built-in such as text-to-speech. Following this, you have eliminated 70% of all Linux distros that fail this.

Also, this distro needs an easy and consistent way of installing development libraries or even the latest libraries without it getting in the way of the developer's work. Whether if it breaks the OS or forces the developer to decipher those cryptic error messages via googlgling for each distro / library combination. Well done, you have eliminated some of the default Linux desktop OSes that do just that.

Lastly, what happens when a dev tries to install your software / firmware update? Is it as clean as a Apple software update? Is there a easy way to revert a failed system update without reinstalling the entire OS or trawling through a blog to copy-paste commands for recovery options?

Unfortunately, every Linux distro requires the latter and there is no standard or easy way of rolling back a bad update without typing a few commands. If we have users that do not mind this issue, then a choice of either Pop!_OS or ElementaryOS look like clear winners to me to pre-install onto a standard or high-spec laptop to sell to devs and consumers.

> Now we need to create a new operating system for an optimal user experience.

No, we don't need to do this.

The devs will decide. I don't want Intel ME (I don't even want Intel), nor secure boot, nor EUFI. I don't want the outdated, buggy, possible malware filled "OS" that comes pre-installed on hardware these days -- that is the first thing to go. I already multi-boot a laptop. I do want an easier way to boot Operating Systems.

The only thing that you need to do is to make your hardware 100% compatible with Linux, BSD, Windows and MacOS (if it's possible). I even bet that if you would build machine that is unofficially, dead simple to run MacOS on it (sort of like everybody knows that Thinkpads run great under Linux) you would win a lot of developers.

linux, larger ecosystem, more driver support