Slashdot Mirror
HP To Kill 3000 System After 30 years
James Ots writes "HP have announced that their 30 year old HP3000 series of computers will be joining their calculators on the scrapheap. Which is a shame, because a lot of work has gone into porting unix tools to the platform, and now we'll have to stop and port MPE (the HP3000 OS) tools to unix. Cnet have pre-announced the announcement, and the guys on comp.sys.hp.mpe don't seem too happy. (See also CSL's page on the story)"
Let's replace all these crappy old systems (hardware + software) with something more decent. Replace HP3000 with HP9000, Cobol with C++, and EDI with XML.
I for one, think that updating hardware and software every 30 years should be mandatory. Think of all the time lost to update and maintain that crap!
Nostalgia is not for Teckies!!! (except when it concerns Arcade games ;o)
By the way, I don't want to hear about Unix being far older than Windows... Unix is still being developed actively.
Black holes occur when God divides by zero.
HP3000's look like this.
Unlike HP's excellent and unparalleled line of RPN calculators, perhaps these minicomputers actually do belong on the scapheap. I lost my 48sx in college and I'm brokenhearted that I can't replace it.
If guns kill people, then CmdrTaco's keyboard misspells words.
Was the old HP Computer games books that used HP 3000 basic. (of course the basic many of them were using was a copy of a DEC basic from a outfit started by a young Ivy league dropout.
Another Wild-Eyed CANADIAN.
BTW, what is the deal with them discontinuing the calculators? I always thought RPN was just about the coolest idea ever for calculators, and I have fond memories of having "calculator races" back in high school where we would see who's machine could solve the problem first. Those of us with the sweet HP calculators were always the first to finish. Truely the end of a great product line.
Is your company running tools written by ma
This server line is as old as UNIX for heaven's sake; I think you have to admire HP for keeping it going this long, especially when you consider they're still going to support it for a few more years. I mean, dropping the calculator division may have been a boneheaded move but you've got to give HP some credit here...
On top of evrything else now my company has to figure out how to migrate over 100 3000 systems, with over a terabyte of data and several million lines of code to a new platform.
lots and lots of tapes, and the mother of all perl scripts.
From the article:
In 1972 and 1973, the early versions [...] were temporarily withdrawn from the market because of flaws [...]
Well, that surely goes in the book of records: it took them 28 years until they decided to make the "temporary" "permanent".
Click here or here.
Here's the early history of the HP3000.
According to the FAQ, it rather ran iX, you're right, I may have been confused between my HP3000 and the HP9000 that came soon after.
BTW, HP-UX appeared in 1986
Trolling using another account since 2005.
What tools? I haven't used MPE/xl in 10 years, but I don't remember it having any tools other than file copy (the OS doesn't even support directories if I remmeber correctly) and db schema stuff.
Although I do remember how me and a guy cracked (yes as in warez) a text editor for mpe/xl once. Each 3000 has a serial code that shows up as a read-only environment variable, and a lot of software uses that as a software key. i.e.: if you tried to copy a program to another box, it saw a different serial and said "no, you copyied this". So our hack was to create a slightly different environment variable called HPSUSAM, and store the serial # from the machine we copied the program from. Then we used a binary editor to search through the program for any occurance of "HPSUSAN" and replace with "HPSUSAM". m41nfr4m3 h4>0r1n6 1s 1337.
_______
2B1ASK1
Not to sound harsh, but isn't that the kind of thing you should have thought of 10 years ago? They stopped deleoping these ages ago, so an exit should have been looked at at least as a side project.
Mod point free since 2001
(well, same numbering system so sort-of related to the HP3000 series)
Ahhh... I have fond memories of the HP2000A (later 2000F) system back in the mid-1970's at the University of Saskatchewan. Not really a "system" 'cause as soon as you logged in, you were dropped into a BASIC interpreter!
It is time to start planning a migration strategy. Evidently there are already some tools available to migrate to other platforms but this might be a good chance for your company to take a step back and reexamine the system as a whole - a rewrite might be in order.
Here is the official announcement from HP.
We found out yesterday morning. An HP service rep called one of our supervisors at home to break the news. While I'm not the one who manages these boxes at our site (I've got the UX machines), I do know that they are the most reliable of anything we've got. They just don't go down, and I guess this is bad news for HP. They need stuff to break so they can boost sales and services.
We've still got several critical apps running on MPE, including our 911 software for PD. These things are bulletproof, and I cringe at the thought of the PD folks going out and choosing an NT solution now. I can only hope a decent 911 app for UX exists.
I worked with a couple of these systems during my last year of college. I can't say that I'm sad to see them go, as I'm all for "better" (subjective, I know) technology. The box and MPE OS were very stable and they provided me with some very valuable learning experiences early on. Backups using those big reel tapes. Ahhh yes! Those were the days.
So, yeah, onward and upward, but it's still nice to look back now and again. We should definitely remember where we came from.
Co-founder and designer at Music Nearby: http://musicnearby.com
Compaq logo? Yeah, HP and Compaq are *planning* a merger, but Compaq is the one being absorbed AND it hasn't happened yet. Someone get a clue.
It appears to be more of the recent flushing-out of whatever is perceived as 'old' by the HP mgmt, including the 'old' HP concept of engineering things to run forever if needed. Is frequent downtime supposed to be a 21st century concept???
And, fwiw, I like HP/UX ... I get a lot fewer calls than the NT guys get ;)
I'd have a personalized plate on my car, but "toxic bachelor" won't fit into 7 letters.
I bet they blamed this on 9/11 too. Oh well, I just felt like burning off some extra karma. being at the cap isn't very fun unless you can post useless drivel at +2.
Hammer of Truth
The HP3000 was the first computer I actually had any "control" over. My high school had a 3rd one with a room full of terminals that they used to teach a Pascal course (if you can imagine). Right after I graduated they were all replaced with PS/2 Model 30's -- which could not have been any where near as much fun as:
...
...well... it's been close to 15 years now since then, it's hard to believe HP was still signing people up for new ones.
1.) Learning how much fun the "down" command was as a cheap prank.
2.) Sending messages from the consoles to newer newbies that their terminal was about to explode.
3.) Mystery Mansion
4.) The Land of Warp -- and I don't care what Adventureland says (http://www.lysator.liu.se/adventure/), Warp was weeks of fun.
Of course, when the AC went out at my school, the nicely cooled server room was a favorite place. Oh yeah -- and I think I still owe my school $376.58 for the service call when I downed the console. It seemed clever at the time.
At least I wasn't the one who hit the Emergency Stop button
Now I suppose where the "kinda" comes in is
The Linux community could really take advantage of this opportunity to score with a killer app for businesses, a HP 3000 Emulator. I know that my company would love to migrate to all of their HP 3000 programs to another solution where they would still have rock-solid reliability and now have commodity hardware prices. This could bring about a true business need for Linux support services and basically bring the motherlode of cash for Linux programmers.
Just think of it, there are thousands of big companies using the HP3000 looking for a solution over the next 5 years (when HP ends support). HP will probably try some god-awful ports to the 9000 series, but if it's not broke, just emulate it. After all, millions of man hours have been invested in getting those programs to handle mission-critical applications.
When someone writes this, let me know... my company has a large pile of cash ready for them.
Once I managed to write a chat program that used the message command, but the only time I seriously tried to use it, I couldn't get the guy on the other side to understand the concept of typing in his session ID instead of his logon ID.
This was also the machine on which I first discovered Crowther & Wood's Adventure. Somewhere in a box in storage I still have the printouts of my sessions on it.
Then the school got this stupid TI refrigerator-looking mini which crashed whenever someone turned off a terminal. But I never got to mess with that one.
--
"Open source is good." - Steve Jobs
"Open source is evil." - Microsoft
As someone who learned how to program on an HP3000 *Series I* (showing my age here), I can't help but feel bad about the decision, logical though it might be. New 3000s (based on PA-RISC hardware shared with the 9000) have been sold primarily as an upgrade path for existing users for quite a while. Apparently, those users (which paid the bills at HP for many years) are (finally) starting to dry up.
My career was made by these machines, although I saw the writing on the wall quite a while back and moved on. I worked for a number of companies that used 3000's (and probably still do in some form or fashion) including a long stint as a 3000 field software engineer with HP itself.
The system aged as gracefully as any computer in history, and was based on boring old dependability, much like the company itself used to be. Between this, the instrument/medical division (now Agilent) and calculators, it feels a little like the heart of the company has been removed.
I was fortunate enough to see the very first HP inkjet (in a little case that the Boise division guy practically handcuffed to his wrist), but had no idea how big it would end up being to the company.
I know there is little room for sentimentality in the computer world, but I have just as strong nostalgic feelings for these old beasties as any vintage video game. They are certainly deserving of respect.
If Linux is around 30 years from now, I think many of you (us) would have some sad feelings if the last copy were being deleted. Even if it was being replaced with something "better".
Should I burn the MPE source code fiche, in tribute?
Smilodon
V V
What is it with titles those days. Everyone one is trying to have catchy once but sometimes they go over board. So lets examine this one: "HP To Kill 3000 System After 30 years" at first glance, this suggests that:
1) HP is about to commit a horrible crime: "kill"
2) There are exactly 3000 unites to be killed: "3000 System"
3) HP will do it 30 years from now: "After 30 years"
Karma stuck at 50? Add 2-5 inches.. err.. 2-5x Karmas Count to your pen1es.. err.. Karma all naturally and private
I'm shedding real tears over this.
Today, MPE has web services, ethernet support, and all the other modern trappings... except instability.
My MPE system maintained uptime in the YEARS... regularly... the OS never failed. Once in a long while (every couple of years) a 10 year old drive would fail & we'd have to deal with it. Because we never bothered to upgrade from the HPIB drives to SCSI, hot swap wasn't an option. But... I will note... it is said that you could upgrade the kernel on these w/o ever rebooting.
If vendors made systems as stable as this today, the world would not know what to do with itself.
-jbn
(Anyone in DC interested in doing a wake / memorial service?)
I worked for Bradmark, Inc. (http://www.bradmark.com) developing database repair and restructuring tools, and it was really interesting work. Sure the user interface was old, but the kind of code I was writing let me get down under the skin of the OS every now and then, and despite what most people here have been saying, MPE was a nice OS, and had features that I have yet to encounter on unix OSes.
File locking for one. I'm sorry, but the unix notion of locking a file is a joke. "I'll create this here lock file, and then other programs that see the lock will know not to open my file. I sure hope the other programs agree to play nice."
Give me a break. In MPE the locking mechanism is built into the file system, and is enforced by the OS. It is easy to build complex locks like "lock bytes 7643-8126 for exclusive write access" and then other programs can do whatever they want with the other parts of the file, and they can read the locked part, but only you can write. *Very* useful for databases.
Another thing the 3000s excel at is stability. I can honestly say that in the 4 years I worked at Bradmark, the only time our development machine ever had any instability was when we ran a beta version of the OS one time for some testing. I once saw an hp3000 ad that actually advertised their machines as having "99.999%" uptime. They had no worries about false advertising, because it wasn't false.
And on the rare occasions when something does go wrong, these machines are designed from the get go to recover gracefully without user intervention. In addition to their external UPS, each machine has an internal battery. This battery isn't for maintaining main power, rather it just maintains RAM, for up to 8 hours or more. When main power is restored the system does a self diagnostic, rolls back in disk IO that had been interrupted, reconnects to any dumb terminals (widely used when the 3000 was first designed), and restarts all programs! If you had a system where all users connect through terminals, then you could sit there and watch all of those terminals come back to life with their programs running exactly where they were when power failed.
Now that's reliability folks!
Someone I worked with told me his favorite 3000 war story: there was a brief power failure in his building during the middle of the day, but power came back on fairly quickly. At 5:00pm the 3000 sysadmins all made a point of walking by the computer room and saying things like "Gee did the power go out?" for the benefit of the Unix admins who were still checking their filesystems and trying to recover their machines.
If you're out there Chris: "Hi!" *waves*
Just like they did with Y2k, they'll do with this. Everyone knew about Y2k for several decades before they started worrying about it. What makes you think that many businesses will connect the pain they had over that with their delay in dealing with the problem and not do so now?
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
Man, I don't think I'll ever forget:
build myfile rec=-72,3,f,ASCII
I learned FORTRAN and BASIC on an HP3000. It's the machine I used to play through Zork for the first time.
I also remember how we had to sneak around the administrators to use our chat program, since they all thought that computers were strictly for class work.
Now that computers are probably used for communication more than anything else, I wonder if they feel a bit sheepish?
-jcr
The only title of honor that a tyrant can grant is "Enemy of the State."
Actually, HP-UX first appeared in 1982 or so on the HP9000 platform (series 500, Focus chipset, 32-bit CISC machine customed-designed by HP). A different version appeared on the 9000 series 200, 68000-based workstation (later replaced by the series 300). HP-UX 1.0 refers the first version on HP-PA (now called PA-RISC).
And, of course, there's the old joke: "If Hewlett-Packard had been named Packard-Hewlett, what would they have called HP-UX?"
Actually, they're still developing them now, and will be releasing new enhancements until 2003.
The profits of doom?
Damn, buying evil geniuses in a nutshell has finally paid off!
From the CSL page:
A very strange thing happened on Wednesday, November 14th. At 11:00 am Pacific Time, hewlett-packard announced that they had decided to discontinue sales and support of the hp e3000 platform. Winston Prather, General Manager, hp e3000 Business, made the announcement with Jim Murphy, General Manager, hp Server Support at his side. [bold added]
The thing is, I'm typing this at 9:49 AM Pacific time.
Curious.
Steve M
Oh, those games are classic... anyone remember Minotaur, and Wumpus?
Does anyone know if the games are available anywhere in a format that will run on Linux, I'd love to get a set...
- Have a picture
I hate those things, let them die a slow painful death at the bottom of the Atlantic. Use them to dredge a new shipping channel. Teach explosives training to new recruits in the Army with them. Use them as obstacles in automotive crash tests. But whatever you do make sure that some back-assed takes forever to upgrade corporation (like the one I work for) cant find them to use them.
If the Compaq/HP merger goes through, I suspect that VMS will end up on the heap too. MPE was the pro-vms people's rallying point that HP might keep VMS alive after the merge.
For systems beyond a certain age, this is just a dumb idea, especially on a large scale. You'd get more performance if you invested the time and money necessary to keep them going in a few cheap current systems.
Many applications don't need any more performance than they already have. What is a dumb idea is to port reliable applications off of reliable computers onto new systems when there is no overriding cost or performance concern driving the effort.
I don't know how Microsoft and Intel have managed to convince people that they need a ghz-class machine to do word processing...
It turned 30 and it's crystal turned red.
Time for Carousel.
So if HP was saying the 3000 was a viable datacenter product last year, it is perfectly reasonable for a customer not to start replacement planning. If HP wants to stay in the data center market, that is.
sPh
I said when the spun off Agilent "HP is dead" - HP was and intrumentation company FIRST, when they spun that divison off I said it was over. I wonder if Agilent can survive on it's own, and if they can, should they BUY the HP name and calculator division when HP liquidates?
-- 73 de KG2V For the Children - RKBA! "You are what you do when it counts" - the Masso
It was a topic of discussion in my undergrad classes in the early 80's and showed up in books as early as 1979 in our library. They said that using 2 digits for dates was a problem then and that it shouldn't be done- ever.
People knew about the Y2k "problem" two decades before the drop-dead date. Whether or not businesses were listening is something completely different- why spend all this time and money "fixing" something that's not broke (as in, it's working right now...), right?
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
Not to mention that most of these systems will be replaced with ones having a more "northwestern" feel :-(.
That is all.
they're in the process of merging. I believe one of the heirs to one of the Hewlett-Packard owners is vigorously opposing it (you'd have to look that up).
A: None. The Universe spins the bulb, and the Zen master merely stays out of the way.
The 3000, on the other hand - closed, proprietary, not the most flexible and capable by today's OS standards - is more and more a niche product. Even if it is still profitable for HP today, it won't be over the long haul. It is not HP's future.
HP has talked about retiring the 3000 line for years. As I understand it, they've kept the line this long only because of their commitment to customer service. There are a lot of companies (like ours) that rely on the 3000. It will be expensive to replace.
People who are critical of companies still using 3000s, IMHO, are a little lacking in real-world business experience. We recognized long ago that the 3000's life was limited. We haven't put any major new applications on them in years. Unfortunately, we have millions of lines of existing code supporting several critical lines of business. We can't replace that at a whim. It will be extremely expensive.
As just one example, the Y2K remediation effort for one large application was about 24,000 man-hours. Note that this application was already almost Y2K compliant, designed in the beginning to track century information. For most programs, most of this time was simply the overhead of checking out the source code, reviewing it for compliance, and checking it back in. There were thousands of programs to check.
I agree that HP deserves credit for continuing the line for so many years past its prime, and for providing good advance notice about retiring it. The future is open systems. HP "done good" by easing the transition.
I used to work an on MPE system about eight years ago, so please forgive me if my memory is faulty.
I was under the impression that it wasn't just the MPE OS that was part of the stability equation, but also the hardware that HP built for these things - at the time I didn't really pay much attention to the hardware specs as I was just trying to get some software working on that system so I can't give exact details.
It seems to me that "rock-solid reliability at commodity hardware prices" is rather an ironic phrase, and I'm not sure is possible...
However, I think that other than that point you have a great idea! People will need to migrate from these boxes and it would be good to have a Linux upgrade path, at least for the mid-term.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
Bummer. That was what I worked on at my first job many years ago. It was the first computer I used with a CRT, and also the first one on which I played "Adventure".
It had an unusual stack-based architecture, and a very nice close-to-the-hardware programming language call SPL/3000, derived from Algol60.
Under capitalism man exploits man. Under communism it's the other way around.
And, fwiw, I like HP/UX ... I get a lot fewer calls than the NT guys get ;)
:-)
You work in sales?
Having been in a shop which employed a variety of Pr1me systems, there will undoubtably be 3rd party support for these for years to come. Just expect HP's end of development to grind to a halt.
A feeling of having made the same mistake before: Deja Foobar
My client ended up choosing an Intergraph (NT based) solution. The CAD and RMS functionality seemed to be right on target for our requirements, we were extremely impressed by the Intergraph personnel, and management (not me) had decided that NT/2000 was the right technological direction to go in.
Unfortunately they didn't end up moving forward on the implementation until over a year after I stopped working for them. I still don't know if the system has gone live yet.
We're moving from 2 HP9000 servers to 1 HP9000 and 1 HP3000. This dictated by the software we'll be running. I expect all applications will, over the next 5 years be ported to something else.
A feeling of having made the same mistake before: Deja Foobar
I could see this coming a parsec away.
;)
;)
In a previous life I did HP3000 development. Ahhhh the memor^H^H^H^H^Hnightmares...
Yes, the HP3000 hardware and OS (MPE/iX) are supremely stable. However everything is also supremely expensive, and performance isn't very good.
The last few years MPE has desperately been playing catch-up with the modern Unix world. The development tools on the HP3000 are horribly archaic -- much worse than even ancient Unixes. The default native MPE environment doesnt even have a fullscreen text editor! At least you get 'vi' with Unix. The OS was riddled with anachronisms at least as many levels deep as Dante's hell. You think Unix is archaic? You ain't seen MPE, baby. It makes VMS look brand spanking new.
The (relatively) recent attempts to bring HP3000 up to speed didn't really work out that well. Adding a POSIX subsystem was cute, but not terribly useful. POSIX stuff could see everything on the MPE side (files, etc), but MPE applications couldn't easily access POSIX data. In the end it was like having two mutually exclusive OSes on the same box. They could co-exist but couldnt really usefully share data.
The HP3000 filesystem is both a blessing and a curse -- the record oriented filesystem can be extremely cumbersome at times when you're used to the rest of the world dealing with simple streams of bytes. Trying to ship data between HP3000 and the real world can be a real hair-pulling experience. Even Macs don't usually have it as bad.
I pity those companies that bet the farm on HP3000's. They may have several years before support is cut off -- but porting tens of millions of lines of code, much of it SPL (basically a macro assembler), is going to be a herculean effort. In many cases it's going to be easier to just start from scratch.
I guess I'm just glad I got out when I did
Groveton High School, class of '82! When I was a freshman, we still had access to a pair of HP2000's for BASIC and RJE, besides the HP3000. We had two 2645A CRT terminals and one printing terminal. Those were the machines I learned to type on! BTW, IIRC, we were running those terminals at 2400 baud, not 300.
In my time, there was a county-wide group that used to meet for movies every weekend. Was that still going on by '84?
BTW, are you sure about the timing when you got those NEC PC's? I remember having two of those in my senior year, and we were using them to run pascal under CP/M. We also had a handful of Atari 800's, which were wheeled around to various classrooms on carts with these enormous TV monitors.
Which school were you in?
-jcr
The only title of honor that a tyrant can grant is "Enemy of the State."
I used one of these for a decade and yes , they are a super reliable set up.
..what is the point of all that uptime if you can't connect due to other crap getting in the way ?
Unless you tried to use that complete screw up of an application 20/20 . Does anyone else remember that botched attempt at a spreadsheet app ?
I recall 20/20 bringing down the HP3000 at least twice before it was dumped.
What people tend to forget is that the 3000 and the OS were reliable and yes , I can recall instances of my terminal coming 'back up' right where I left off....but the 3000's that are around today are largely accessed via PC terminal apps (reflection etc) thereby exposing reliability to all the vagaries of MS desktops and all the network glitches that come with the 'No dumb terminal' approach.
From an end user perspective anyway
Yes, Warp was loads of fun except when you hit the places where the puzzle was not solvable unless you could run over and ask Rob Lucky or Bill Frolik for hints.
-russ
Don't piss off The Angry Economist
"But... I will note... it is said that you could upgrade the kernel on these w/o ever rebooting."
Errr, no. Upgrading MPE was always a convoluted process, requiring rebooting.
... have been greatly exaggerated.
This is the HP3000 they're talking about. This means the death of MPE/iX (HP3000), not HP-UX (HP9000, enitrely different OS).
gspl compiler for what, i386?
:-)
A lot of SPL code I've seen is selfmodifying code. This means pushing old (pre-parisc) opcodes onto the compatibility mode stack and executing them.
Don't forget that a lot of SPL code depends on intricate details of the compatibility mode linker, too.
In the end, if you were going to do a gspl you'd basically have to end up writing a compatiblity mode VM as well.
Even HP didn't get the CM VM 100% perfect when they went to PA-RISC. CM code still sometimes mysteriously vomits (or maybe its on purpose, part of evil scheme to get you to port your code to native mode
Don't forget the complexity of writing a VM to fully emulate the intricate block-mode oriented terminal I/O... even commercial terminal emulation software like Reflections don't always get this quite right...
I don't think it would be an easy job at all. You could make a bundle off it though, selling it to corporations desperate to keep their HP3K investments afloat...
Anyone in their right mind relegated to maintaining ancient code should take a serious look at the requirements the code is meant to satisfy and ask for money to rewrite and replace that code, if only because the maintenance costs of both the code and the hardware will soon be prohibitive.
"Databases that can easily be in the gigabyte range" are considered micro databases these days. If Excel and VB don't have the capacity, then just about any other DBMS and GUI will.
--Blair
"Y2K won't be the last bug attributable to 'oh this shit'll never run that long...'"