After 47 years, OpenVMS gets a package manager (VSP) | Hacker News Reader

After 47 years, OpenVMS gets a package manager (VSP)

raymii.org

102 pointsjandeboevrie37 comments

It already has a package manager. Emacs.

so politically incorrect

I wonder if someone could have ported pkgsrc to it, create a POSIXish compatibility shim library (maybe there already is one), and, I wonder how much of the denizens lingering in the pkgsrc would have ported over “for free”.

VMS has had posix compatibility since the 90s. You can install gnu tools on it as well.

How many OpenVMS systems are out there in production?

A lot. Problem is that sometimes the hardware is lost due to renovation to the building.

https://www.theregister.com/2001/04/12/missing_novell_server...

I haven't seen one since the early 2000's. That company has since ported their code to Linux and AIX.

We still operate hundreds of legacy (2nd and 3rd gen) systems with Alpha hosts running OpenVMS in airgapped prod environments across the country.

The portfolio of apps we developed/maintain for these legacy systems have all been ported to their modern replacement systems years ago, but we still regularly release updates targeting legacy that maintain feature parity with modern releases.

What’s that like? Fun? Annoying? Interesting? Boring?

All of the above; being a mid-career IC/TL with visibility into actual outcomes in prod has organically matured a deep respect for boring technology---warts and all---especially when cumulative NRE is well into 10-digit territory.

I'll be glad when legacy support hits EOL though, if only for the indirect consequence of unambiguously seperating wheat from chaff as we embrace a new tech stack approaching stability. Although it isn't the case, it sure has felt like we've been operating like cost center baggage these past few years.

Curious how you handle hardware failures for all that old equipment, etc. Do you have a huge depot of spare parts? Ebay? etc

I'm a R&D plebeian; not my job. We have a lifecycle management team that monitors metrics---e.g. system availability, failure rates, spares inventory, obs, etc.---and postures as risk/budget considerations permit.

In the aggregate, these legacy systems are on the right tail of the bathtub curve quickly approaching EOL, hence their modern replacements already being fielded. As legacy systems are decommissioned, reclamation buffers the spares inventory pool. Push come to shove, cannibalization in the field keeps most balls rolling at the cost of diminished availability/throughput.

Based on a sample size of 2 legacy systems in our lab that saw daily use over the past decade, things definitely broke, but I don't recall the production Alpha hosts every giving up the ghost beyond CMOS battery replacement nuisance.

We still have a few support Alphas laying around just in case, but virtualization[1] on commodity bare metal effectively renders them redundant.

[1] https://www.stromasys.com/solution/charon-axp-alpha-virtuali...

Equipment of this time frame was highly reliable. The machines were expensive to purchase and very solidly built. I would be surprised if this fleet of systems deals with a hardware failure a year. Especially if they are kept in a space with cold conditioned air.

It doesn't support my VAX!

"Translate all custom C, BLISS, FORTRAN and DATATRIEVE applications on this VAX into Golang and reconfigure all custom DCL scripts to run in BASH on Linux. Output the results as a bootable disk image for GB10 processors."

yay!

Both openvms users will be ecstatic!

oh my

But the real question is can you add a Firefox ppa? /jk

It’s considerably out of date, but theoretically yes? https://h41379.www4.hpe.com/openvms/products/ips/firefox/

Congratulations on the new package manager. But despite its name, OpenVMS is still actually not open.

> But despite its name, OpenVMS is still actually not open.

This is a very anachronistic complaint - DEC renamed VMS to OpenVMS in 1992, and at the time “open” had nothing to do with “open source” since that phrase wasn’t even coined until 1998.

In 1992, “Open” meant “open standards”/“open systems” - support for industry standards instead of proprietary ones - most importantly POSIX. VMS became OpenVMS when it acquired POSIX compatibility. And other vendors used the same branding-IBM offered mainframe POSIX compatibility under the names OpenEdition (MVS) and OpenExtensions (VM)

> since that phrase wasn’t even coined until 1998.

This idea that Christine Peterson invented the term open source in 1998 is revisionist history. I personally remember the term being used during the early 90's, and here's a Usenet thread from 1990 that backs up my memory:

https://web.archive.org/web/20220213152832/https://www.tech-...

(The relevant text is a little way down in the thread: "BSD's open source policy meant that user developed software could be ported among platforms...")

Heck, here's a Caldera press release from 1996 that argues directly against your point that OpenVMS couldn't have been confused with open source. In the press release, Caldera announce their "OpenDOS" which is, they say, an "OPEN SOURCE CODE MODEL FOR DOS"

https://web.archive.org/web/20200521004332/http://www.xent.c...

I don’t agree it is “revisionist” history. Yes, people used phrases like “open source code” prior to 1998, but without a clear and universally agreed meaning - pre-1998, “open source” could just mean the source code was available under unspecified terms, possibly with significant strings attached (e.g. commercial use requires paid license, no redistribution of modified versions, etc). It was only circa 1998 that “open source” came to clearly mean the Open Source Definition

In fact, in a pre-1998 sense, VMS (“Open” or not) had “open source” in that customers could order source listings. Yes we’d now call that something more like “source available” but that category wasn’t clearly distinguished from “open source” back then

In any event, go read trade magazines from the early 1990s, you’ll find talk of “open systems” in almost every issue, the phrase “open source” rarely if ever occurs

If I remember correctly it was more or less coined as an alternative to "Free Software" which wasn't as industry-compatible. I also remember a lot of discussions (and essays by Stallman) about how Open Source is not Free Software as it lacked the enforcement of freedom that is the GPL.

You are acting like the term open source has a clearer and more agreed upon meaning than it did back then. Perhaps the idea of a more standard definition has grown, but people still use it confusingly today.

current open-source also has all sorts of strings attached depending on the licences -_-. wanna start that debate really?

funny how you move the goal post from 1998 to early 90's.

I built my first linux box in 1996 and yes open-things were largely associated with open licenses already so it is no suprise some companies decided to jump on the openwashing wagon early.

> funny how you move the goal post from 1998 to early 90's.

I didn't "move the goal post from 1998 to early 90's".

We are talking about OpenVMS which was given that name in 1992.

So the "goal post" has been the early 90s all along. The relevance of 1998 here is that it is after the early 90s.

> I built my first linux box in 1996 and yes open-things were largely associated with open licenses already so it is no suprise some companies decided to jump on the openwashing wagon early.

Nobody was "openwashing" in 1992

I agree and remember similar. I wouldn't be surprised if someone else says that "open source" doesn't mean "Open Source" in the "official" (whatever that means) sense. Wikipedia seems to support the revisionist history: https://en.wikipedia.org/wiki/Open_source

It’s not anachronistic to make this complaint about a product still being sold in 2025. The product name is now an anachronism.

Per their wiki -

> The operating system originally called VAX/VMS was renamed to OpenVMS around 1991 to reflect the high degree of support for industry standards such as POSIX.

> This prefix is now generally regarded as silent by OpenVMS users and developers calling the operating system "VMS".

https://wiki.vmssoftware.com/OpenVMS

Yeah, that has always confused me. When I first started playing with Linux and BSD, I would see OpenVMS popping up in my searches, only to find that it wasn't actually open.

I'm sure there's legacy reasons for that, but it always sort of annoyed me because I wanted play with it

It's "open" in the sense of the Open Software Foundation, OPENSTEP, or OpenGL. Back in the pre-CATB days, "open" meant "developed by a consortium rather than a vendor", or perhaps even just "ported to more than one CPU architecture". This latter sense is what is meant in OpenVMS, as it was ported to DEC Alpha CPUs in the mid-90s.

I believe in this case, "open" = supports UNIX/POSIX apis.

And that there was documentation on how it worked.

It was literally a complete bookshelf completely full for the docs.

Always bothered me too. At least POSIX is actually a POS.

Reply on news.ycombinator.com