32 comments

[ 126 ms ] story [ 220 ms ] thread
Emphasis on paid. I do appreciate the business genius of identifying and capturing a market consisting of customers so inept that they can't get off Solaris even if you give them half a century to do it.
> customers so inept that...

I used to think like that until I got to see some legacy systems in action. Sometimes it just makes financial sense to keep paying Larry and avoid a big rewrite.

This, and also Solaris was years ahead in some areas. Hopefully linux will overtake it completely before 2037.
I wonder if bcachefs can do it.
Are there any such advantages that you can't get with illumos?
Support. (Like, official vendor support & hardware support)

It's really nice to type a command and see a CLI representation of where the drives are physically located on your system for example.

All the IllumOS distributions are defunct (SmartOS hasn't seen a release since 2020 and that's the most recent) aside from OpenIndiana, which I haven't looked at in a long time.

*EDIT:* Seems OpenIndiana is not binary compatible due to using glibc over Sun Studios libc. Which might prevent some people from switching.

> All the IllumOS distributions are defunct (SmartOS hasn't seen a release since 2020 and that's the most recent) aside from OpenIndiana, which I haven't looked at in a long time.

Er, what?

Tribblix 0m33 2023-12-14 https://tribblix.blogspot.com/2023/12/changes-in-0m33-prerel...

OmniOS CE r151048 2023-11-06 https://omnios.org/article/r48.html

SmartOS 20240125T000404Z 2024-01-25 https://us-central.manta.mnx.io/Joyent_Dev/public/SmartOS/sm...

noghtly builds are not releases
Where are you seeing nightlies? The OmniOS release notes I linked open with

> On the 6th of November 2023, the OmniOSce Association has released a new stable version of OmniOS

And the tribblix release doesn't look like a nightly.

The SmartOS one... might be? But AFAICT it's just rolling release.

Solaris may have had a handful of helpful features but even by the late 1990s it was obviously inferior in numerous ways to Linux and BSDs. One of the most obvious manifestations of how slow it was was the overwhelming latency of fork, orders of magnitude slower than its free competitors and the reason its ecosystem needed hacked up threads libraries. The system was sprinkled with surprise complexity traps that could kill you in production, including the fact that its TCP receive path was O(N) in the number of IP addresses associated with a given network interface, meaning if you tried to hang an entire subnet off 1 port the system would effectively hang. In 1998 the people I worked with could not run away from Sun quickly enough. As soon as we could port anything to FreeBSD, we did. The writing was on the wall even then.
By 2006, Nokia was still mostly a HP-UX and Solaris shop on the networking side, and CERN still had quite a few Solaris boxes, with Scientific Linux project alongside Fermilabs slowly taking off in 2003.

Not everyone was racing to jump out of UNIX proper during the late-1990's.

The article specifically discusses Solaris as a gleaming success for web startups in the 1990s. I am here to tell you that as a member of that scene, I would have burned Solaris at the stake if it had a suitable physical manifestation.
And I am here to say, I had more fun with Solaris during dotcom wave that ever had, or will, with GNU/Linux.

Same applies to a couple of big UNIX names, with a proper integrated experience.

Pity about that Google Android torpedo.

Yep, same reason why a lot of businesses also still run AS/400
I do like how this seems to stop curiously short of 2038.
This was my first thought too
Per the official support document:

> Sustaining Support Ends: Indefinite

* https://www.oracle.com/us/assets/lifetime-support-hardware-3...

If you're willing to cut a chequing, they're willing to sell you a support contract.

Solaris has supported a 64-bit time_t since (at least) 2010:

> The 64-bit system interfaces are inherently more capable than some of their 32-bit equivalents. Application programmers concerned about year 2038 problems (when 32-bit time_t runs out of time) can use the 64-bit time_t. While 2038 seems a long way off, applications that do computations concerning future events, such as mortgages, might require the expanded time capability.

* https://docs.oracle.com/cd/E19455-01/806-0477/6j9r2e2ah/inde...

can vs must, or even should.

2038 will be a problem, because using 64 bit time was optional.

Binary compatibility was kept, but 2038 will have some chaos, and Oracle won't care because support will have already ended.

> support will have already ended.

But the support is "indefinite", meaning it doesn't haven't an end date for the customers on that tier.

I have a couple of pizza boxes under a bed in the spare room. My SO wants me to throw them out, I say in a low voice "you'll thank me when the balloon goes up ..."
Can you expand on what you mean by this? I researched online for this reference and consulted with some colleagues while discussing this topic but none of them could figure out what you mean by when the balloon goes up.
In general, when the SHTF, or things rapidly deteriorate. When the war finally starts. Etc.
Quite hard to believe when typing the phrase into Goggle brings up the dictionary definition as the #1 result.
Because you'd need like 10 solar panels to run one of those, compared to like 2 for a modern system? ;)
Modern systems are already BIOS-compromised by the state, powering them up would instantly summon the "termination drones".
Then these must be some old pizza boxes to exist before governments wanted to slipstream backdoors into hardware
Not sure how that'd work after shit hits the fan badly, as there probably isn't going to be a working residential network for it to alert anyone over. ;)
Brings back memories. I started my first business providing Weblogic hosting on a e450 in a cabinet we bought from Enron, running off of a DSL line. Fun times!
I never used it myself, but had to transport an e450. I got an Ultra 10 out of it.
(comment deleted)