5 ms·
"I've described Bob a little in "The Early History of Smalltalk" that I wrote for the ACM in 1993" That was an amazing read. So much connects now. I'm going to
by nickpsecurity 10y ago
"I've described Bob a little in "The Early History of Smalltalk" that I wrote for the ACM in 1993"
That was an amazing read. So much connects now. I'm going to have to think on it as it's a lot to take in for one morning. I do spot some things to immediately reference.
"They seemed and do seem: magical."
Well, when I read the paper I linked, it does look like so much then-unheard-of stuff compressed into one, short document that magic is apt description. It led to B5000. Not just that, though. Your Smalltalk history showed it also helped shape OOP and your work. A Roger Schell interview showed INFOSEC basically started with Anderson: a Burroughs engineer that looked at people, software, and hardware to find systemic, security risk. Their methods are similar to Barton's w/ different perspective. I bet Anderson's thinking traces to it & Burroughs B5000. So, Barton seemingly out of thin air hits all kinds of critical points in concept space of IT, helps put them into a demo (B5000), and its ripple effects help create OOP (indirectly) and INFOSEC (directly). Fair to call it magic as most people were more narrow and incremental with nowhere near as much impact.
" I should also note that Bob was a mathematician, thought like a mathematician, and designed like a mathematician"
That makes a lot of sense. His work resembles high-assurance a bit in that he worked from clear specs of what it had to do to primitives that could do it to right combination of pieces. It's like he was formulating and solving equations for high-level programming itself. It's why I originally skipped the paper because I wasn't mathematical enough to read some of it. Then, your HN comment said Barton, I had one paper with that name, and now all kinds of light bulbs are turning on. :)
Speaking of people doing magic, I landed on Margaret Hamilton's team a year or two ago:
https://en.wikipedia.org/wiki/Margaret_Hamilton_%28scientist%29 https://en.wikipedia.org/wiki/Margaret_Hamilton_%28scientist...
http://htius.com/Articles/articles.htm http://htius.com/Articles/articles.htm
The "Lessons learned from Apollo" paper suggests they were similarly working in a vacuum around same time as Burroughs stuff was coming out. Invented all sorts of engineering and software principles in the process. Are you aware of any others work that went into that or any influence of such work on your side of the fence? As in, was her team's work... especially person-high stack of correct code... as independent and ground-breaking as it appears? Or building on prior work in a way we just haven't seen?
Just curious as I've tied INFOSEC, Burroughs, and your work but not sure how her work or legacy fit into the larger picture. Outside safety-critical systems that is.
" the later Flex Machine as well"
Is the Flex Machine with tagged hardware and Ten15 VM the same as yours or different? Yours mentioned an "aerospace" company. In any case, I wish I had more details on both as they seem like an extension on Burroughs thinking that sort of appeared and then disappeared. I think a HW-accelerated Ten15 could be valuable today. Even if it was an imperative-style RISC + one customized for functional stuff with a clean IPC or functional call mechanisms to integrate anything written for both.
https://en.wikipedia.org/wiki/Flex_machine https://en.wikipedia.org/wiki/Flex_machine
"via mixing microcode with fixed functions."
I keep recommending groups doing RISC-V, etc to use microcode instead of fixed. The reason being old work showed microcoded processors could be repurposed to do many things. The microcode was also used for speed boosts and atomic execution of arbitrary, instruction sequences. Do you second the recommendation of keeping microcode in CPU's wherever possible?
" Part of the need for the hybrid was to tinker with various kinds of storage schemes, formats, garbage collection, swapping, etc., that were not well understood enough to be put directly into hardware"
Here's the two leading candidates that are the successors of Burroughs and capability-oriented systems:
http://www.crash-safe.org/papers.html http://www.crash-safe.org/papers.html
https://www.cl.cam.ac.uk/research/security/ctsrd/cheri/ https://www.cl.cam.ac.uk/research/security/ctsrd/cheri/
SAFE combines tagged architecture, functional programming, and a device for arbitrary, policy expression and enforcement. It's clean-slate with strong properties. CHERI implements a capability architecture that allows fine-grained enforcement and C compatibility. Runs CheriBSD, a FreeBSD port. I see CHERI as an interim solution with SAFE going in direction of more, ideal system. I think they can be mixed. Yet, I think CHERI might not be thorough enough and SAFE is more complex than needed. I thought a subset of SAFE could run an Oberon System port with instant practicality.
So, here's what lines I'm thinking along that I'd like to run by you. I started with i432 successor, the i960. The manual has nice reference of its capabilities. Amazing compromise between an i432 and a RISC.
https://en.wikipedia.org/wiki/Intel_i960 https://en.wikipedia.org/wiki/Intel_i960
What it had was common RISC stuff you can run anything with, MMU, error handling of common errors, and object descriptors for OOP constructions or POLA enforcement. The other thing it had was speed despite all that. So, start with something like that for RISC-V or OpenSPARC. Implement, a la B5000 or SAFE, some specific, tiny-cost checks for basic primitives: pointer/array bounds-checks; stack protection if it has one; code vs data tag (or Itanium-style read/write/execute per word or page); labeling & protection of pointers themselves. These checks run in parallel in tiny HW as CPU runs so essentially no overhead. IOMMU auto-manages tagging of incoming or outgoing I/O data to fit the model. Any failure takes CPU to exception handler with access to relevant data. Optional, dedicated HW for concurrent GC possibly built-in to MMU like some LISP machines. Implemented either clean-slate or with configurable, open CPU like Gaisler's Leon3.
It seems those minimal, primitive protections knock out 90+% of the problems. Functional can be mapped to them. Extra error handling or object descriptors, like in i960, do much more while avoiding heavy-weight constructions. I think this is a nice start. I have primitives for safe concurrency, too. What do you think is absolute minimum, safe primitives on top of that we need for OOP like a safe Smalltalk? I don't do dynamic, late binding, or any of that currently. I don't know if it needs special consideration in HW or if prior HW above could handle it with compiler doing all that. Anything come to mind?
And thanks for the time and helping me connect dots in IT history. :)
- alankay1 10y agoI'll try to reply to this in sections. One way to think of "magic" is height of leap. Newton had something before him, but his leap was cosmic. I don't think we've had a Newton yet in computing, but we have had some cosmic leaps. Just to pick a few: Sketchpad, Lisp, and the B5000 ideas all had something before them, but their leaps were enormous. In the "Early History of Smalltalk" I mentioned that I had seen and partly learned the B5000 while in the Air Force ca 1963. I say "partly" because I didn't understand all the implications until a few years later in grad school. For example, I did understand what was great about the byte codes and stack mechanisms, but it took until later to really grok that the machine was organized so that environments and access were -granted-, and that the protected tagged pointers -- Descriptors -- could not be forged, and that the actual code contained no addresses. (This was the real start of "capability-style protection.) Also, the multi-tasking OS -- written in an Extended Algol -- was not just the first OS in a higher level language, but was more profound than Algol (it was "Simula" before Simula). This was done extremely well because -- and most people don't realize this -- the B5000 had multiple CPUs! More in a next note ...
- alankay1 10y agoI've been a huge fan of Margaret Hamilton for a long time. Her story is in many ways what can be wonderful about great design and creation by really smart and able people, and so disappointing about our not-quite-a-field. (And if you want a tough computer to code and make a great system for, look no further than the Apollo Guidance Computer!) I think that a mainframe computer like the B5000 (or more conventional mainframe) was just too much machinery back then to overlap much with the needs of Apollo). Hamilton realized that a really good notion of "module" would help tremendously with integrity, and that the OS should have an active "overlord" to dynamically assess and deal with real-time conditions and needs for resources. I think this was just amazingly brilliant work for that time (and any time). NASA did give her a major award (but where was the ACM?). I've never had the pleasure of talking with her, but I have a feeling that NASA could also have done something to encourage her ideas to be more communicated and put out into the open.
- alankay1 10y ago