SoylentNews is people
SoylentNews
OS news brings us the news that HP-UX reached the end of its life on December 31st:
It's 31 December 2025 today, the last day of the year, but it also happens to mark the end of support for the last and final version of one of my favourite operating systems: HP-UX. Today is the day HPE puts the final nail in the coffin of their long-running UNIX operating system, marking the end of another vestige of the heyday of the commercial UNIX variants, a reign ended by cheap x86 hardware and the increasing popularisation of Linux.
HP-UX' versioning is a bit of a convoluted mess for those not in the know, but the versions that matter are all part of the HP-UX 11i family. HP-UX 11i v1 and v2 (also known as 11.11 and 11.23, respectively) have been out of support for exactly a decade now, while HP-UX 11i v3 (also known as 11.31) is the version whose support ends today. To further complicate matters, like 11i v2, HP-UX 11i v3 supports two hardware platforms: HP 9000 (PA-RISC) and HP Integrity (Intel Itanium). Support for the HP-UX 11i v3 variant for HP 9000 ended exactly four years ago, and today marks the end of support for HP-UX 11i v3 for HP Integrity.
And that's all she wrote.
HP-UX 11i v1 was the last PA-RISC version of the operating system to officially support workstations, with 11i v2 only supporting Itanium workstations. There are some rumblings online that 11i v2 will still work just fine on PA-RISC workstations, but I have not yet tried this out. My c8000 also has a ton of other random software on it, of course, and only yesterday I discovered that the most recent release of sudo configures, compiles, and installs from source just fine on it. Sadly, a ton of other modern open source code does not run on it, considering the slightly outdated toolchain on HP-UX and few people willing and/or able to add special workarounds for such an obscure platform.
Over the past few years, I've been trying to get into contact with HPE about the state of HP-UX' patches, software, and drivers, which are slowly but surely disappearing from the web. A decent chunk is archived on various websites, but a lot of it isn't, which is a real shame. Most patches from 2009 onwards are unavailable, various software packages and programs for HP-UX are lost to time, HP-UX installation discs and ISOs later than 2006-2009 are not available anywhere, and everything that is available is only available via non-sanctioned means, if you know what I mean. Sadly, I never managed to get into contact with anyone at HPE, and my concerns about HP-UX preservation seem to have fallen on deaf ears. With the end-of-life date now here, I'm deeply concerned even more will go missing, and the odds of making the already missing stuff available are only decreasing.
I've come to accept that very few people seem to hold any love for or special attachment to HP-UX, and that very few people care as much about its preservation as I do. HP-UX doesn't carry the movie star status of IRIX, nor the benefits of being available as both open source and on commodity hardware as Solaris, so far fewer people have any experience with it or have developed a fondness for it. HP-UX didn't star in a Steven Spielberg blockbuster, it didn't leave behind influential technologies like ZFS. Despite being supported up until today, it's mostly forgotten – and not even HPE itself seems to care.
And that makes me sad.
When you raise your glasses tonight to mark the end of 2025 and welcome the new year, spare a thought for the UNIX everyone forgot still exists. I know I will.
Did you work with HP-UX? What did you think of it? How does it compare to more modern OSes? More widely, can we still learn things from older software, and are they worth archiving as historical items?
(Score: 5, Interesting) by zocalo on Sunday January 04, @07:35PM (2 children)
And, of course, once we had figured out a few privilege level hacks, we transposed a couple of letters in the OS name..
UNIX? They're not even circumcised! Savages!
(Score: 3, Interesting) by Anonymous Coward on Sunday January 04, @10:13PM (1 child)
A college I worked for had a single HP-UX box in the library serving their card catalog. Agreed HP-UX was not the worst commercial unix. Of those that I've used, AIX definitely held the spot for worst, WTF?! and JFC! But, HP-UX also had "little quirks" like, mounting a CDROM with Rockridge (aka unix) extensions, would cause a kernel panic, so standard procedure was to mount CD on Solaris/Linux, and export it over NFS to HP.
It used to be said [...] that AIX looks like one space alien discovered Unix, and described it to another different space alien who then implemented AIX. But their universal translators were broken and they'd had to gesture a lot.
-- Paul Tomblin
(Score: 2) by driverless on Monday January 05, @05:41AM
That's like saying "Gonorrhea was not the worst STD". I'd actually forgotten about AIX when I made my post further down, but I still rate DGUX as the worst I've ever worked with. AIX was actually not that bad once you grokked the IBM-specific ways some things were done, it worked, just in an IBM-ish way. So I'd say anything-else > PHUX > AIX > DGUX. All of them sucked, that's just different levels of suckiness.
(Score: 3, Funny) by bloodnok on Sunday January 04, @08:48PM (2 children)
He always lamented that the company wasn't called Packard Hewlett.
__
The major
(Score: 2) by driverless on Monday January 05, @05:34AM
We always referred to it as PH-UX. On a scale of DGUX to Ultrix it was definitely well towards the DGUX end. I'm sure some of my former colleagues will be dancing little jigs of despair that it's been retired.
(Score: 0) by Anonymous Coward on Wednesday January 07, @12:35PM
They'd have to change this too: https://www.hewlettpackardhistory.com/item/turning-hp-upside-down/ [hewlettpackardhistory.com]
(Score: 4, Interesting) by jelizondo on Sunday January 04, @09:17PM
Back in the day, the company I worked for bought two HP 832 for different offices, one of them for our site.
It was a real computer (*), with networking support (coax cable only, if memory serves right) for connecting PCs but you could still attach dumb terminals if needed.
I might be mistaken, but to me it was the start of real, distributed computing. I remember enjoying very much working with HP-UX on the 832.
(*) Well, compared to other computers we had. The first computer I ever saw was, I think, an IBM System 36, I might be wrong about the model, but definitively it was IBM hardware (this happened almost 50 years ago, so the memory is murky) and I was only allowed to look at the blinkenlights. It was fed punched cards and it spit more punched cards as output.
Then I got me a Commodore Vic 20 with 4k(!) of RAM and later a Commodore 64, but fascinating as they were, they were toys really.
My boss bought an Atari 800 without asking me and decided to use to automate some task or other. It took me a long while to find out why the cassette tape storage wasn't working; turns out that when the recording to tape sequence started, it would output a high pitched noise that scrambled whatever data you were trying to save. The trick was to start the recording sequence and not send data for a couple of seconds, then everything worked as advertised. Piece of shit.
So indeed to me, it the HP 832 with HP-UX was a real computer.
(Score: 4, Informative) by isj on Sunday January 04, @09:22PM
I made programs for HP-UX from 1999 to 2015. 32-bit PA-RISC, 64-bit PA-RISC, 64-bit Itanium. I found that things mostly just worked.
There were some oddities, though. 32-bit PA-RISC supported shared libraries but used a different API (shl_load/shl_unload).
The PA-RISC CPUs were interesting beasts with some wild features. Eg. 1 clock cycle load from L1 cache because it was virtually-mapped and didn't have to wait for TLB lookup. Some revisions could speculate 53 instructions ahead. And delays slots after a jump could have a jump in them with well-defined semantics.
The Itanium version had fewer oddities. It just worked.
(Score: 5, Interesting) by psa on Sunday January 04, @10:26PM
I was senior HP-UX admin for a time. It was not as refined as AIX, not as innovative as Solaris, not as flexible as (pre-systemd) linux. It worked, most of the time, and changed little. Patch sets were a mess and it was ridiculously expensive to license in enterprise configurations. Most of our trouble came from the itanium hardware, and it was endlessly problematic. I always had at least one critical system with suspected hardware issues, and I spent endless hours trying to prove them to HP support, but that was mostly mucking about in the firmware. Our local HP tech was great, and he had practically had his own workspace onsite, but he could only do so much to get the company to listen.
I really didn't have any trouble with the OS, but I couldn't justify the astronomical cost. My life got better when I moved on, so I can't really mourn the OS. As far as I'm concerned, HP-UX died when they discontinued PA-RISC, and it just took them a long time to figure out the obvious.
(Score: 4, Insightful) by SomeGuy on Monday January 05, @02:52AM (1 child)
It is worth archiving as much as possible. But one person or even a small group of people can only do so much. Much software has disappeared completely. Electronically distributed-only may disappear the moment it stops being used. CDs and floppies may be sitting in some attic, but it takes time and motivation to do anything with them. Without the hardware or even an emulator, how does one even test it? Outside of DOS/Windows/Mac, every platform is sort of its on 'island". It takes a good bit of time and knowledge to get in to it. And business stuff that doesn't run games.... yea, eyes are going to glaze over.
Heck, I've spent a good bit of today trying to reconcile multiple crufty wares copies of a particular DOS application - all copies are missing parts. And eBeh has mostly dried up for such things. It all takes time, effort, and money.
(Score: 1) by shrewdsheep on Monday January 05, @11:41AM
My stack of old CDs has almost dried up indeed. The very least everyone can do is to upload an image to archive.org. While not helpful at the moment it might enable future software enthusiasts/archeologists and adds a little to overall preservation. It takes time, effort, and money but doesn't have to take much of those (in proportion).
(Score: 3, Interesting) by bzipitidoo on Monday January 05, @08:18AM
For one job, I had to make a Perl and C application run on HP-UX. When I got it all working, they threw a further curveball at me by insisting it run on an older machine with an older version of HP-UX that didn't have some of the library functions of the then current HP-UX that I had used. One of these was the lack of variable length file names. So, to make that work, I used a fixed length equal to the length of the longest file name, and padded names with leading slashes as needed: "///////tmp/subdir/filename.pl". Gross, but hey, it worked.
(Score: 2) by VLM on Monday January 05, @03:00PM
It's a problem for retrocomputing that people LOVE stockpiling and sharing software and docs that they enjoyed but stuff that was kind of a PITA ends up lost in the breeze.
People really enjoyed DEC operating systems so there's tons of stuff out there despite legal issues (in some cases) even Ultrix is floating around.
But then you mention Apollo which got acquired by HP at the end of the 80s and I can reminisce about my life back then but I really have no desire to run HP/Apollo (or HP/UX) from the early 90s again. The uni engineering building had literal classrooms full of Apollo workstations; I don't miss using them at all. I don't remember Domain/OS at all (it is not in the HP/UX family tree but is very close to it) all I remember (possibly incorrectly) is it's the only unix-alike that had parts of the OS written in Pascal.
(Score: 2) by turgid on Monday January 05, @09:42PM
The migration is complete. You can turn off your Unix servers now.
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].