alblue

I've written up a more detailed piece at InfoQ:

https://www.infoq.com/news/2021/01/java-turns-25/

The Goldman office is not fully staffed any more, and many bankers are leaving for Europe. The euro market has gone already, and others aren’t far behind.

https://www.bloomberg.com/news/articles/2021-01-15/brexit-is...

https://www.bloomberg.com/news/articles/2021-01-18/london-fi...

The other aspect you’ve failed to notice is that many companies are eschewing offices now. They’ve proven they can work without them; Docker, for example, is now going fully remote.

To be honest, the above comment is exactly why Brexit happened: a refusal to see the bleeding obvious while chest-thumping and saying “we are the greatest”.

It’s why trade is basically avoiding the UK and new sea routes have opened up between Ireland And Europe — because the UK has made trade inefficient. No amount of chest thumping is going to change that dynamic, and England will be poorer because of it.

Reality doesn’t care about Brexit ideology.

The last “word” in the title needs updating to have some space in it so it breaks properly on mobile devices.

Yup, my presentation was back in March 2020 and the M1 came out later in the year — sooner than I was expecting, TBH; I thought that it was a couple of years out when I said it :-)

The problem is that leap seconds might have made sense when time keeping was accurate at the millisecond level but now that time can be measured at the nanoseconds level (at the speed of Computing, essentially) such changes are to big and cause problems, especially if they are negative. Instead, applying by smearing the time over a period (see Google’s approach) is a less dangerous way of handling the situation, but now you’ve got computers that might not agree on time exactly.

As a result it is easier to abolish leap seconds and just live with the fact that in 30,000 years time that noon might not be directly overhead.

Swatch tried to do that by creating Swatch Internet Time, which was a time zone less decimal time system. As you can guess, this didn’t end well or take over the world.

https://en.m.wikipedia.org/wiki/Swatch_Internet_Time

Time zones exist so that (approximately) the sun is overhead at noon. If you ask anyone what time zone the world should operate on, they’ll agree, provided that they use your time zone. Otherwise you could make everyone use GMT/UTC instead, and I can’t see America doing that when most of the country can’t count beyond 12 on a clock.

There are some counties which use a single time zone by dictat, like China, but all that happens is people adjust their working days such that the sun is up at lunchtime, so you don’t gain any advantages by having everyone on the same time zone because now you need to know what individual areas working hours are so you can communicate.

These days more problems are caused by leap seconds, which is the difference between GMT and UTC, and I expect we will get rid of those soon as well.

I have never seen that term be used in this way before, nor can I find any references on that. Do you have links to be able to justify this claim?

See also https://twitter.com/TwitterSafety/status/1346970433917222912

It can be mitigated somewhat if you have pcid enabled for the processors, so that a flush can be process specific instead of cpu-wide. There’s still a performance hit of course but you’re going to lose some locality with a context switch anyway.

Sort of; Intel and AMD are stuck with the variable width instruction isa that exists due to historical evolution. To do something different you need a new isa.

Intel tried this with Itanium a while back and failed because it is difficult to get software developers to target a new isa and provide compilers and compiled code for everything unless you use a translation layer.

Apple is one step ahead here because their compilers already supported ARM isa (because iPhones use them) and had both the OS and apps ready to go from day one of availability.

They also had translation technology that would allow mutating x86_64 code to ARM64 code so that old apps would (on the whole) run acceptably fast on the new chip.

To do the latter properly, Apple had to create a special mode to run the arm chip with total store order for memory writes, which is not standard on arm. (It would be a lot slower if they didn’t have that when running Rosetta translated code.)

So both the OS being available, and the OS influencing the ARM tweaks (eg TSO) could they pull it off.

They also have the position that they build hardware that uses those chips so can mass produce - and in fact, replace - existing hardware.

Each of these things could be done in isolation by Intel/Windows/Apps but it would be difficult to do all three.

Even getting JavaScript maths in a special instruction was difficult enough on Intel, and that was something of benefit to any browser.

My guess is you’ll see Intel and AMD offering Arm chips in the near future, as both AWS (graviton) and Apple have shown the way to a new ARM future.

This isn’t specific to the M1 but I tap about cache lines in my last QCon presentation (where I also suggested that a 128b cache line wasn’t far away):

https://www.infoq.com/presentations/microarchitecture-modern...

However the speed benefits come from a much larger L1 cache and the fact that the ram is in the same chip which will reduce latency that is the benefit for most of it.

The program (instruction) cache is also a lot bigger and has the advantage that as a fixed size isa can be much wider in execution than in x86 but that’s unlikely to be of benefit here, other than perhaps slightly in terms of queuing up multiple outstanding loads.

To be clear here, it fetches two cache lines but it doesn’t put the second in exclusive state until it’s written to; the unit of granularity is still 64b. In a scanning read mode you will see the benefit but you won’t see the contention on writes. (The contention will come from subsequent reads on that cache line though)

The big problem with the nightingale hospitals is that there were no staff to run them, and hospitals are understaffed as it is with people of sick or isolating due to the governments approach of vaccinating their voters first ahead of medical staff.

Even if they were used in the first wave it was in the expectation that any hospitals sending patients there would also send staff to look after them, which was never realistic.

You can hire a hall and stick a bed in it, but if there’s no one to look after the sick person in it they don’t have much of a chance.

Interesting paper, but I noticed a couple of typos:

* ntel in the conclusion

* x96 in the main body

(in case the paper authors read this commentary)

Indiana Jones and the Crystal Skull has what turned out to be aliens and an alien spaceship.

Time to emigrate from the U.K. to a more enlightened country.

What countries would you go to as an English expat for a better life? Where can you change a useless English citizenship for another?

Not really. Under the covers, it has always been an object database where the objects key is the hash of that object, and other than blob, tree, commit and signed tag they haven’t really changed since then.

The protocol has become a bit smarter, by allowing there to be additional options passed for different communications options, and there is a slow but steady progress of moving from sha1 to sha256 for the hash function, but that isn’t rolled out yet completely.

Different tools have been added over time but the internals are pretty much the same as they’ve always been.

That only works when your SVN server is in the same computer as you; the fact that SVN has to reach out to get additional data for the history is what kills it.

With git on the other hand, all the file storage is local to you machine regardless of which other remote git servers you may be using.

In addition, git blame doesn’t have to go through all history; the logical git storage model is object based on hash, so it’s really a set of queries for the contents of hash values that is used. Yes, it walks through commits, but at the lightweight metadata of trees which are akin to navigating a file system.

Longer exposure, but the lights in the room were dimmed as well to allow it to show up. The brightness was pretty limited. No filter used other than the iPhone native camera in low light conditions.

I took a photo of it when I was at the computer history museum last time:

https://twitter.com/alblue/status/1340463649597616128

Al Lowe is on Twitter at @allowe if you’re interested, and has a site at allowe.com.

I can also recommend the Ken Williams book, by the way.

Typo in title — should be “designated”, not “designed”. Dang, can you fix?

The linked video on the piece introduces the term “job queue” by means of programs, after being punched onto paper tape, are hung up on hooks for subsequent execution by the committer operation when the prior job completes.

https://youtu.be/6v4Juzn10gM

This is possibly the first time that a “job queue” has been recorded on video (1951) in relation to computer programs, and since the computer was single threaded and could only run one computer program at a time, it was necessary for the programs to wait in line and be manually processed in turn.

Yeah, my mistake - have fixed. Thanks for the feedback!

Early days, but Hypervisor.framework patches are being built that allow QEMU to run an Arm build of Linux (and Windows) at native speed.

https://twitter.com/jonmasters/status/1333555759737663489

The reality is that some of the drivers for the M1 chip will never arrive, but that if you want to run Linux (or Windows) Arm code then you can do so soon.

Thanks for the feedback. Will update the article accordingly.

You are of course correct that the CA was announced then, but the news article was also about the fact that Let’s Encrypt had just turned five years old on the day that the article was published, and was a look back at what they have achieved over the last five years as well as looking to the future.

It was also noteworthy that they are moving to ECDSA and have reduced the length of their certificate names to save bytes (packets) in the initial TLS connection.

When I was a kid decades, we used to go sledding in the snow approximately every Christmas.

These days snow is such a rarity that my kids have hardly ever seen snow in this country. They’ve gone sledding once or twice in their lives.

Furthermore, the idea of a white Christmas is largely a byproduct of the Charles Dickens era in the late 1800s when he saw snow every Christmas and even skating on the Thames when it froze over.

https://en.wikipedia.org/wiki/River_Thames_frost_fairs

The bcc-tools ship a Clang/LLVM compiler to compile and build the BPF programs, which takes additional resources and adds a requirement for kernel headers to be installed on the machine.

The libbpf-tools uses a compiler at build time (not run time) to generate a bytecode (portable) program that need no additional dependencies on the server other than the libbpf, and so requires less setup and resources at runtime to install the filter.

The M1 is an Arm+ in the sense that it supports a stronger memory model option that plain Arm doesn’t, and it’s this feature that is used by Apple’s OS and the Rosetta translation engine that gives the performance boost for x86 translated programs.

Of course the Rosetta isn’t needed by Linux and assuming that all software is recompiled for Arm natively then you probably wouldn’t need it, but I suspect that is what the GP was referring to.