Hacker News new | ask | show | jobs
by sour-taste 2241 days ago
Then why doesn't everyone use Debian?
8 comments

Debian has its own set of problems. Like the Chromium package maintainer deciding unilaterally several years ago that installing extensions remotely shouldn't be allowed and gated that standard functionality behind a command ling flag.

There was zero documentation on the change and the error you received attempting to install an extension was basically "operation failed". I discovered the cause only because there was an open bug about it on the Debian bug tracker where the maintainer refused to acknowledge the problem. Eventually, sane minds prevailed and that stupid patch was reverted.

So, unfortunately - you'll end up having to deal with people that refuse to look at things from the user's perspective no matter what distro you use.

And yes - I'm still bitter. :-/

There is also the ffmpeg kerfuffle a few years back when Debian decided to replace the ffmpeg package with an incompatible and inferior fork. You can imagine the amount of confusion that ensued.

That said, I think Debian's occasional messups are far less egregious and damaging than Ubuntu's though.

There was also that time many, many years ago they decided to remove the ability to load binary firmware blobs for things like network cards in the kernel because of an interesting interpretation of the GPL. :-/
Maybe when they tought that none-free drivers opt-in was a good idea, also old kernel / packages.

LTS on Ubuntu was always better than Debian, I mean saying that Ubuntu is just a repackage of Debian is very short sighted, especially on the security side, Canonical security team is top notch.

> Canonical security team is top notch.

They need to, for a long time, Ubuntu has been shipped with EOL Kernel versions https://ubuntu.com/kernel/lifecycle

Canonical is at a disadvantage here because Red Hat directly employs or has significant established relationships with many of the people who make kernel release decisions. Plus, RHEL is the de facto standard enterprise distribution, which means any decision the kernel community makes regarding what they believe "enterprise" requires will often be a reflection of Red Hat's plans.

But what benefits Red Hat in the enterprise world is to their detriment in the consumer world. There's a reason the Debian/Ubuntu package ecosystem is richer and more featureful than RPM, and this is why Ubuntu dominates in the container space--because almost any piece of software that one could expect to have been packaged has been packaged as a .deb and already exists in the default package archives. I can't count the number of times I couldn't find an RPM--certainly not in the default repositories (RHEL, CentOS, or even Fedora), but not even in the third-party community repositories. And those that do exist are of lesser quality than the comparable .deb, for various reasons. (That is, the long-tail of packages is of higher quality for Debian.)

By pushing Snap, Canonical is definitely going astray. Ubuntu's competitive advantage is the Debian package ecosystem. Both Canonical and Red Hat seem to underestimate the role and importance of their respective packaging ecosystems. How many projects to revolutionize or replace RPM/Yum/whatever at Red Hat have crashed and burned? Many, though it's hard to count because half-way through they often realize what they're trying to do is functionally or even technically impossible (as with their aborted 2017 plans for RPM package streams), and scale things back to iterative improvements.

Containers are a security nightmare, and pretty much the only reason to pay Canonical and Red Hat licensing fees is for security and bug fix maintenance of their package archives. On our large Kubernetes clusters at work there are thousands of open CVEs for the containers that are being run, and we'll have to boil the oceans to get them all updated, let alone keep them updated. But updating packages is as simple as an apt-get/yum upgrade[1], and rarely do you have to worry about anything breaking, especially relative to the pain that updating containers regularly brings.

[1] If the container uses Ubuntu, Red Hat, etc you can sometimes just rebuild the container to get the newer packages. But that assumes you control the container image. Most containers come from third-party, decentralized sources (that's the point!). But Docker Hub doesn't cajole and coordinate container owners to update their crappy images. It's no substitute for the orchestration of people that are traditional package repositories.

> also old kernel / packages.

So, LTS/stable.

Ubuntu has been marketed as a beginner friendly distro, with communities easily accessible using a Google search (thinking about the likes of askubuntu, omgubuntu, ...). So I'd say online presence and beginner-friendliness.
Similarly, Mint is also one of the most popular distros, and its marketing, at least for a long time, has been that it's even more user-friendly than Ubuntu.
Marketing?
Third-party support.
Lots of software out there with Ubuntu PPAs.
Debian did not send free CDs by mail
Did a large numbers of people use those? Legitimate question—it always struck me as a cool initiative for a very small number of people, but only that.

I'd expect most people tech-savvy enough to install Ubuntu would also have a decent enough internet to download a ~700mb file.

Back in the day, not much of my country, or even the US, had particularly fast internet. Nor did everybody have a disc burner in the days before USB booting being supported by the majority of computers' firmware.
If my memory serves correctly, this was 2004/2005, around the time I was discovering my home burnt CDs and DVDs were going bad.

This was also around the time I would often brick my primary (only) workstation for whatever reason. Having a properly mastered Live CD was super useful.

I would order at least two with every release cycle for a few years at least.

Thankfully, I saw the light early with Ubuntu-server, and stayed with Debian. Ubuntu-desktop makes for a good enough live / recovery / troubleshooting environment, but not sure I’d use it for anything more.

The free LiveCDs were great marketing. As a teenage computer geek, it was way easier to convince casual computer users to try it out when I could lend them a nicely printed CD. And it looked way better than handing them a sketchy CD-R with some marker scribbles on it.
> I'd expect most people tech-savvy enough to install Ubuntu

we installed ubuntu with friends in freakin junior high school. it's not like it's rocket science...

OTOH the village where I grew up as a child only got DSL > 512k around 2008 iirc.

Checked a bit, the figure I could find was an ubuntu employee estimating the figure to half a million in 2004, so at the beginning of the programme -> https://ubuntuforums.org/showthread.php?t=1691&page=2&p=3255...

So yeah, lot of people got those.

I ordered a few back in the day - the sleeve design was really nicely done.
Because it's popular it gets 3rd party support. So people use it making it popular. Vicious/virtuous circle depending on whether you're looking at it from above or below.

Just like windows Shuttleworth admires and seeks emulate so very much.

Sigh. ok then let me explain that last comment.

It's called a network externality. Microsoft achieved this with piracy based market penetration of dos and it worked great for them and they've built on it ever since.

Shuttleworth understood the importance of getting established as being popular when he launched and would press install disks, as many as you asked for, and ship them to you at his own cost. As one example of nakedly going after market share and spending resource to do so.

Bug #1 in the ubu bug tracker is literally "windows is the most popular os."

Separate to the marketing, which is worth discussing on this site because some of us actually care about what works and why so wish to discuss it, let's talk engineering decisions.

Between 2 choices that are technically about equal. Choose the one that is more popular. Many feet trample more bugs. Better support. More likely to be around after $time_period. If it needs to work with something else, the managers of the something else project will likely suppor the more popular first etc. Obviously popularity is not the only concern but it really does count for something, ignore it as a dimension in your decision process at your peril. I use linux, and ubuntu as it happens on this laptop. I'm aware windows and osx are more popular and that popularity makes certain things easier. For /my/ purposes and to /my/ taste linux and ubuntu are worth paying that cost to have installed here and I'm very comfortable with that decision.

Just quietly, perhaps people who don't care for business decisions and engineering decisions are on the wrong website? There's plenty of places "boosters" can go to do that.