VMS on Raspberry Pi 5

I would agree, however, everyone in the company was well over retirement age. We offered to help them set up their own shop. I guess they found it an opportunity to switch to WEENDOZE.

A decade or so ago, a program called 'monit' handled monitoring and restarting of daemons quite nicely. However, I don't know whether it's still maintained.

monit 5.3.3 released Feb 2023 is in FreeBSD ports. So yes it is still maintained and quite popular.

System V’s packages were, what, 1990 or so? So nearer a third of its current age I would say. I’m not even sure they were the first.

Be that as it may the fact that some of them are in principle portable doesn’t change the fact that nobody has ever much tried to use any particular one outside its home platform. So I think they form a reasonable counterexample to the idea of a common goal of portability of system-level components.

Well, yes, there’s practical reasons why it happened like that, but they aren’t insurmountable: we have common extension APIs in many other contexts, but apparently nobody was ever motivated to do it for drivers, despite the obvious advantages.

SMF and launchd are both counterexamples AFAIK. A lot of the original Unix design decisions were fairly novel at the time too but many of them have proved to be good ones.

The lack of portability of systemd can hardly be an objection to it if nobody in the platforms that it can’t be ported to is interested in adopting it anyway.

There’s still nothing there that demonstrates Linux ever had a goal of portable system-level tools. I get that you think it should do, but wanting isn’t getting.

LOL!

Who will bell the cat?

GNU always has as its primary goal that of providing portable system-level tools. For a long time, Linux was the kernel and GNU provided most of the actual environment in the distribution.

However, at some point the momentum built up until "Linux" as a concept took off, and then everybody kind of forgot about GNU and the GNU environment and philosophy... then they forgot about the Unix environment and philosophy... now it is just a mess.

--scott

That list was what Amazon use not what AWS offers to customers - that list is longer.

Very true.

When big data is actually big then clusters with thousands of nodes is the only way to handle the data.

Arne

Engineers don't do philosophy usually (I am the exception) except ' Perfection is the enemy of good enough'

"An engineer can for five bob what any damned fool can do for a quid"

I suspect that Richard Stallman trying to get everyone to call it GNU-Linux didn't help.

Maybe; it depends on how you define "package." Two things happened in relatively close temporal proximity that contributed here: 1) we had vendors shipping closed-source Unix distributions coupled with their hardware; and 2) we started to have networks of Unix machines at a single site.

With these, it became pretty clear that we'd need some kind of principled way to build deploy and maintain third-party software packages and configuration data; a number of different mechanisms were invented, some ad hoc and based on convention, others based on the idea of bundling up "packages" (which could help OEMs and third-party software vendors distribute their software). Work for this really started heating up in the mid- to late-80s.

- Dan C.

Indeed from "The cathedral and the bazaar" to the cathedral and the bizarre.

A philosophical statement

There is reality. This is what physicists and engineers deal with.

Then there is how reality is described, what people *think* it is, what people think it *ought* to be, and how people *think* it can be changed.

This is the world of the Left wing armchair intellectual and the ArtStudent™.

It can for all practical purposes be safely ignored. It is in the end mere 'Bandar Log-ic', the inane chatterings of the Monkey People...

"We all say it, so it must be true".

Engineers deal with models that they hope correspond to reality but don't always (cf. Tacoma Narrows bridge). Physicists attempt to build models that correspond to reality but don't always (cf. contradictions between general relativity and quantum theory). Then there are chaotic systems and emergent phenomena which so far have proven extremely resistant to modelling.

Nobody can prove that reality exists (cf. solipsism etc) but it is the most useful assumption we can make. Even with that assumption nobody can prove that reality is logically consistent - it might all be the whim of God you can't prove that it isn't - but it is the most useful assumption we can make and it hasn't (yet) been disproved.

The core unspoken (usually) theorems of physics are that reality exists and that it is logically consistent. If you can prove either of them I expect you're due a Nobel and an FRS at least.

That is ahistorical. The primary goal has always been production of a complete free operating system, not just component parts that could run elsewhere. See either the GNU manifesto or indeed the current project standards, which still categorise Linux as secondary to the GNU kernel.

Portability of some components (e.g. language implementations) has obvious advantages for that goal, at least until it’s been realized, but that’s a means to an end, and it’s nowhere stated as a primary goal.

Linux took off because it was a usable free platform that was easy to download and run, and easy to contribute to. At the point it got going the GNU kernel wasn’t ready (still isn’t) and BSD was mired in lawsuits. Change either of those and history would have been quite different.

No. In general, I think Stallman has done more harm than good to the free software movement in recent years.

--scott

Because in recent years it has not been about software.

When RMS talk about software then some adore him and some hate him.

But when he talks about anything else then everybody hates him.

Arne

Stallman was never worth a crap on software. We had Open Source and Free Software long before he came along to try and screw it up.

bill

Ok, there ARE apps like Google that NEED what are essentially "infinitely scalable" solutions.

However the poster WAS right about something ... the more you go that way the more vague and unmanagible the 'security' issues become. Docker/Kubernetes are NOT as secure as their authors like to claim and indeed any super-"spread out" solution will also suffer issues. It's a TRADE-OFF.

Big banks/biz, govt/mil, need to err on the side of security. Google can err on the side of high-volume.