The recommendations are often very good, for example Ousterhouts A Philosophy of Software Design, but seem to be on software development in general, not actually software architecture in particular.
For that, I would recommend the classic texts, such as Software Architecture: Perspectives on an Emerging Discipline (Shaw/Garlan) and really anything you can find by Mary Shaw. Including more recent papers that explore why the field of software architecture did not go the way they foresaw, for example Myths and Mythconceptions: What Does It Mean to Be a Programming Language, Anyhow? or Revisiting Abstractions for Software Architecture and Tools to Support Them
More practically: look at why Unix pipes and filters and REST are successful, and where they fall down and why. Hexagonal architecture is also key.
And a plug for my own contribution, linking software architecture with metaobject protocols as a new foundation for programming languages and programming: Beyond Procedure Calls as Component Glue: Connectors Deserve Metaclass Status. An answer to Mary Shaw's Procedure Calls Are the Assembly Language of Software Interconnection: Connectors Deserve First-Class Status.
Answering the question: if procedure calls are the assembly language, what might a high level language look like? And also maybe that software architecture might have a brighter and more practical future ahead of itself.
> if procedure calls are the assembly language, what might a high level language look like?
I’m not well versed in PLT and SWE tooling, but isn’t that the base concept around lambda calculus, LISP, APL, clojure, TCL,…? You only need a few data structures and types, a small collection of fundamental functions, then you compose them.
If there’s one thing that I like about Lisp is that more complex type are always opaque (especially the ones from FFI). I would love to see CLOS implementation for a C-like language, where when you define a struct, what you get is a standard collection of functions.
Next year for first Plasma in the SPARC reactor, which is supposed to yield net positive fusion energy.
Early 2030s for the commercial reactor feeding fusion-powered electricity into the grid. For which they (a) have an electricity purchase contract with Google[1] and (b) just applied for a grid connection[2].
Which of 2027 and "early 2030s" is "30 years away, in your humble opinion?
That joke is really, really past its sell-by-date. Something like 30 years past...
There many technical changes which will have to be overcome for fusion to:
1. Produce long term stable plasma. SPARC should demonstrate this.
Commonwealth Fusion Systems are confident that with the magnet properties of SPARC and the conservative parameters of the SPARC plasma they can demonstrate this.
2. Produce the prototype of a fusion power plant, the ARC reactor.
Commonwealth Fusion Systems has to scale up the SPARC reactor. They have to not only design, but also build and test, reactor vessel which would allow useful heat recovery and tritium breeding. Tritium is hydrogen, very small molecule that has tendency do leak through smallest cracks. It can also cause hydrogen embrittlement.
3. Produce energy from fusion on commercial scale and at competitive prices.
Reactor vessel is exposed to very strong magnetic field and as result large forces. The neutrons produced from D-T plasma have energy about 13 MeV and they cause faster material degradation than in fission power plants. It's possible that frequent replacements of the reactor vessel will be needed. The reactor vessel is neutron activated low-level radioactive waste, which has to be properly disposed of. The beryllium used in some designs is quite rare (world's annual production of beryllium is 220 tons). All this issues affect the economy of fusion power plant.
I confident that all this non-trivial material and engineering changes will be solved and commercial fusion power plants will exist in 50-100 years.
For quick introduction I recommend watching: "International Colloquia #26: Fusion Reactor First Wall Cooling"
Actually the way China builds nuclear reactors is very typical of the way western countries built reactors back when we still did it well: standardize, build a bunch in overlapping batches. Keep building.
That shouldn't be surprising, because they learned it from us.
We stopped doing it that way because we effectively stopped building.
China is building enough reactors that they can do this with several standardized designs. Which is smart.
The EPR has basically failed, so in the west we currently have 3 standardized generation III(+) designs: The Westinghouse AP-1000, the South Korean APR-1400 and the Japanese ABWR.
Of these, both the ABWR and the APR-1400 have been built quickly and cheaply, with the ABWR holding the record for fastest build times: under 4 years.
The AP-1000 had some very rough initial builds, because the design wasn't actually finished and it turned out what they had "finished" wasn't actually buildable. Ooops. These issues appear to have been ironed out, and a lot of countries are betting on the AP-1000: the US, Poland, China, and Ukraine. Turkey, Slovakia and Bulgaria have also expressed interest.
The EPR is essentially dead, with only the UK wanting to build two more UK-EPRs at Sitwell-C. Hopefully the EPR2 will be better, what I've seen of the specs suggests it has a good chance.
Anyway, one point I want to come back to is the "keep building".
This is actually crucial, and one of the reasons many western projects in recent years went so badly. We had forgotten how to build, no longer building a bunch in overlapping bunches, but single units decades apart.
And there comes to rub: in order to "keep building", you have to build slowly. Slow is smooth and smooth fast my guitar teacher used to say. The French built out far too quickly, constructing 55 reactors in just 15 years. Then they were done. Nothing to build until that initial batch wears out. Reactors last a long time, easily 60-80 years.
Ooops.
The key to this comes from queueing theory, Little's Law:
L = ƛW
"the long-term average number of customers (L) in a stationary system is equal to the long-term average effective arrival rate (λ) multiplied by the average time that a customer spends in the system (W)"
So if you have a desired fleet size of 80 units and they last 80 years, you should be completing 1 unit per year. China is currently permitting 15 per year. If they keep that up throughout the construction phase, this would imply a steady-state fleet size of 1200 reactors.
That's a lot of reactors.
If you build more quickly, you won't be in steady state. Of course you can still do better than going full tilt and then stopping, smoothly modulating the build-rate.
For France, this would have meant a fleet size of 320 reactors at the rate they were going. Alternatively, the build rate for the fleet size they have would have been around one reactor every two years.
Something to keep in mind for the "not a lot of nuclear is being built"-crowd.
Rejecting nuclear waste site is an easy and almost cost-free way of garnering browny points with the part of your electorate that has been indoctrinated into massive radiophobia.
It is almost cost-free because in reality, nuclear waste is so low in quantity and so easy/unproblematic to store "temporarily" that it just isn't a real problem. Politicians know this. So they can play this game.
And once pressure builds enough you dig a hole in the ground like you always could have and like the Fins just did and start storing.
For wind this makes a little bit of sense, but for solar "whenever I want to deliver" is largely once per day as regular as clockwork for several hours and that means you can bridge with BESS.
You can already see it in charts, initially BESS shifts some of that peak midday sun energy to evening usage where it's worth more to us, but gradually competition drives down that evening price and so the BESS cuts deep into the night chasing those higher prices. It's most exaggerated in Australia today, where the reason the power is relatively cheap when you wake up before dawn isn't that somehow coal is less expensive at night - it is because much of that is solar power from yesterday and if they don't sell it to you now for whatever price they can get they've wasted a whole cycle, 'cos the sun, with free power, is coming up like it or not.
> once per day as regular as clockwork for several hours and that means you can bridge with BESS.
If you live in Arizona or in tropical climate maybe. For anybody else it is bullshit.
Solar production fell to few percents of its peak when the sky is covered.
Many European regions can spent multiple weeks during Winter with the sky entirely covered.
BESS is nowhere near the capacity required to even go pass a single day. And it is unlikely to change even over the next 10y.
So hoping to run entirely on Solar + BESS for a multi week Dunkelflaute is living in dreamland, no reality.
What happens in practice is that country like Germany will need to have a backup Gaz that matches its peak consumption in Winter if they want to go full renewable.
The other option is to throw the problem on your neighbours with interconnects. This is what Germany does with mainly Norway, Sweden and France. And this is not a sustainable solution.
Well yes, Russia is Russia. The RBMK with its well-known design flaws could never have been certified/built in the West. Yes, these design flaws were well-known long before the Chernobyl accident.
In his posthumously published memoirs, Valery Legasov, the First Deputy Director of the Kurchatov Institute of Atomic Energy, revealed that the institute's scientists had long known that the RBMK had significant design flaws. Legasov's suicide in 1988, following frustrated attempts to promote nuclear and industrial safety reform, caused shockwaves throughout the scientific community.
However, the units that are still operating were modified after the accident to remove at least a few of the elements of the accident chain that made the reactor inherently unsafe.
Still no containment, and still not anywhere close to the requirements for Western reactors, but they seem to be operating reasonably safely.
The subject was modifications made to reactor types that had major accidents, and specifically the RBMK-1000.
Last I checked, the VVER-440 is not an RBMK-1000, did not explode at Chernobyl, and has not had a major accident.
"The two VVER-440 units in Loviisa, Finland have containment buildings that fulfil Western safety standards.
A typical design feature of nuclear reactors is layered safety barriers preventing escape of radioactive material. VVER reactors have three layers:
1. Fuel rods: the hermetic zirconium alloy (Zircaloy) cladding around the uranium oxide sintered ceramic fuel pellets provides a barrier resistant to heat and high pressure.
2. Reactor pressure vessel wall: a massive steel shell encases the whole fuel assembly and primary coolant hermetically.
3. Reactor building: a concrete containment building that encases the whole first circuit is strong enough to resist the pressure surge a breach in the first circuit would cause.
Compared to the RBMK reactors – the type involved in the Chernobyl disaster – the VVER uses an inherently safer design because the coolant is also the moderator, and by nature of its design has a negative void coefficient like all PWRs. It does not have the graphite-moderated RBMK's risk of increased reactivity and large power transients in the event of a loss of coolant accident. The RBMK reactors were also constructed without containment structures on grounds of cost due to their size; the VVER core is considerably smaller"
It also appears that the VVER-440 does have a containment building, and older variants that had problems with said containment building were forced to shut down.
"One of the earliest versions of the VVER-type, the VVER-440, manifested certain problems with its containment building design. As the V-230 and older models were from the outset not built to resist a design-critical large pipe break, the manufacturer added, with the newer V-213 model, a so called Bubble condenser tower that – with its additional volume and a number of water layers – aims to suppress the forces of rapidly escaping steam without the onset of a containment-leak. As a consequence, all member-countries[citation needed] with plants of the VVER-440 V-230 type, as well as older types, were forced by the politicians of the European Union to shut them down permanently. Because of this, the Bohunice Nuclear Power Plant had to close two reactors and the Kozloduy Nuclear Power Plant had to close four. Whereas in the case of the Greifswald Nuclear Power Plant, the German regulatory body had already made the same decision in the wake of the fall of the Berlin Wall."
Nothing man-made is "completely safe". No such thing.
However, nuclear energy is the safest form of energy production we have.
By far.
And that includes Chernobyl and Fukushima.
People overestimate the danger from nuclear energy by incredible amounts.
That doesn't mean that close exposure to a running nuclear reactor won't kill you in short order. That's why we build these things with shielding. A lot of other things will kill you in short order if exposed to them: cars/trains in motion, for example.
For that, I would recommend the classic texts, such as Software Architecture: Perspectives on an Emerging Discipline (Shaw/Garlan) and really anything you can find by Mary Shaw. Including more recent papers that explore why the field of software architecture did not go the way they foresaw, for example Myths and Mythconceptions: What Does It Mean to Be a Programming Language, Anyhow? or Revisiting Abstractions for Software Architecture and Tools to Support Them
More practically: look at why Unix pipes and filters and REST are successful, and where they fall down and why. Hexagonal architecture is also key.
And a plug for my own contribution, linking software architecture with metaobject protocols as a new foundation for programming languages and programming: Beyond Procedure Calls as Component Glue: Connectors Deserve Metaclass Status. An answer to Mary Shaw's Procedure Calls Are the Assembly Language of Software Interconnection: Connectors Deserve First-Class Status.
Answering the question: if procedure calls are the assembly language, what might a high level language look like? And also maybe that software architecture might have a brighter and more practical future ahead of itself.
reply