* Posts by coconuthead

79 posts • joined 8 Jan 2017

Page:

This'll make you feel old: Uni compsci favourite Pascal hits the big five-oh this year

coconuthead

interview with Niklaus Wirth

This interview with Wirth from sometime after 2000 popped up in my Youtube recommendations:

https://www.youtube.com/watch?v=BJIqHIYSDrk

He taks about the origins and history of Pascal, then about Modula-2 and the Lilith workstation, Oberon, and finally his retirement project.

coconuthead

Re: Modula-2

Indeed, it shouldn't have compiled, which may lead the programmer to assume the counter must in fact be signed. This was someone else's program I was porting (from a 60-bit word-aligned machine!) and the declaration was probably hundreds of lines back, as tended to happen with Pascal languages since declarations could only appear at the start of procedures or globally. After that head-banging moment, I didn't fall for it again.

But compile-time checks are IMO not enough here, because of underflow wrapping. Consider a loop with complex logic where sometimes you decrement the counter by 1 and sometimes by 2, for valid reasons. And suppose you get it wrong. With a signed counter at -1, your loop will run terminate even if it runs once too often. With an unsigned one, it wraps to a large positive number, will hang, and perhaps scribble all over memory. I know which I'd rather debug. You can argue the mistake shouldn't have been made anyway, but then why are you using a "safe" language?

Question: in Pascal, is [0..65535] an unsigned 16 bit or a signed 32 bit number? I guess that was the problem Wirth was trying to solve with CARDINAL.

coconuthead

Re: Berkeley Pascal

It was a very weak Pascal, though, lacking things like modules and type coercion that were in the DEC Pascal that ran on the same hardware with their proprietary VMS operating system. It puzzled me enormously why computer science departments were so enamoured of Unix when the programming languages were so very far behind those available on other operating systems.

(BSD Pascal was definitely a good choice for your project, against the ones you mentioned. Young people might not know that C compilers of the era often issued no type errors at all, especially in function arguments.)

coconuthead

Re: Modula-2

In the mid-1980s I had a job where I was simultaneously using Pascal, Modula-2, assembler and some C, all on VAX/VMS. Some others in the team were using Modula-2 quite heavily on embedded networking devices, and my main use of it on the VAX side was tooling support for them.

Two things about the language I hated: required upper case and the pervasive use of unsigned integers, which Modula-2 called CARDINAL. Unsigned counters are error-prone when counting towards zero: if you write < 0 in your loop test, it will never terminate, so you get a hang. The tools written in it that showed up for me to port were also surprisingly nonportable, with the packed/unpacked mess hanging on from Pascal.

I don't remember whether it was a requirement of the Modula-2 language, but I seem to remember screens of import statements, one identifier at a time, the equivalent of C++'s "using MyNamespace::myIdent" rather than the get-it-done "using namespace MyNamespace;". Meanwhile, VAX Pascal did allow importing everything in the module.

(There were 2 VAX Pascal compilers, the original one and the new, incompatible one. It was the new one I used. Modula-2 would have been an improvement over the old one.)

Apple promises third, no, fourth, er, fifth time's a charm when it comes to macOS Catalina: 10.15.5 now out

coconuthead

the login screen says it all

There it confronts you, an aerial shot of the eponymous rocky island - as if taken from the cockpit of a plane about to crash into it.

Record-breaking Aussie boffins send 44.2 terabits a second screaming down 75km of fiber from single chip

coconuthead

Re: Only part of the problem

It is suspiciously close to exactly 4 times the distances between the two campuses of Monash they said they tested between. (It is 18km from Clayton to the city.)

ALGOL 60 at 60: The greatest computer language you've never used and grandaddy of the programming family tree

coconuthead

"Call by name".

coconuthead

Algol 68 is not ALGOL 60

Algol 68 is basically a different language from Algol 60—and the Algol 60 designers, most notably Djisktra, were less than impressed by it, and along with others on the committee issued a "minority report" disowning Algol 68. Syntax aside,and leaving out the object-oriented stuff, Algol 68 actually bears a distinct resemblance to C++. Algol 60, on the other hand, begat Pascal. So really they are different languages sharing part of a name.

I never got to use Algol 68 because it was difficult to write a compiler for and nowhere I worked had one. Burroughs Algol (a variant of ALGOL 60), on the other hand, was available. Time pressure, and perhaps a desire not to be seen to know too much about the "old iron", meant I never did write any Burroughs Algol. That was perhaps my loss. I don't think I missed anything by not writing Algol 68.

BTW Burroughs Algol was implemented in hardware. It was a stack machine with display registers and hardware support for resizing arrays.

DBA locked in police-guarded COVID-19-quarantine hotel for the last week shares his story with The Register

coconuthead

Re: Perfect example...

Australia started requiring all arrivals from overseas to self-isolate at home for 14 days on 15 March, almost exactly when the couple would have started their holiday. Qantas announced on 19 March that they would be shutting down most of their international operation at the end of the month. A total ban on Australians travelling overseas was issued on 20 March and at the same time Australians overseas were advised to return home immediately. The lockdown started on 22 March.

Sorry, no sympathy. Especially since I am awaiting a test result and am currently required to self-isolation (and yes, the police may come to check I am here). I could have only caught it from food shopping or exercising in the local park.

Europe publishes draft rules for coronavirus contact-tracing app development, on a relaxed schedule

coconuthead

iOS update would be a blocker on the Apple–Google scheme

From a quick look around the Apple–Google scheme, it appears it will require an iOS update on the iPhone.

There are probably a lot of people whose phones can run iOS 13, but have been avoiding it because of reports of serious bugs: in the Wifi hotspot, speakerphone, Safari and other important functions. Having a reliable phone is more important than privacy issues at the moment. If I do end up in hospital my smartphone will be pretty important to me!

Apple are quite capable of blocking, and do block, security updates to iOS 12 for any phone which can run iOS 13. There is no reason to think they will do things any differently here.

The Singapore app, on the other hand, just runs on what is there.

Official: Office 365 Personal, Home axed next month... and replaced by Microsoft 365 cloud subscriptions

coconuthead

leaping into a new paradigm in product nomenclature

Surely they change their product names so often it wouldn't have been *too* much trouble to make it Microsoft 366 for just a year?

Forget toilet roll, bandwidth is the new ration: Amazon, YouTube also degrade video in Europe to keep 'net running amid coronavirus crunch

coconuthead

Not all telework is valuable

User 1: some marketing lowlife or organisation (wo)man videoconferencing.

User 2: a child, who cannot see their friends because school is closed, cannot see their nan because old age homes are closed to visitors, and is possibly terrified based on the adults around them, whose mind could be taken off things by some streamed video.

I'm afraid I don't see User 1 as more worthy of bandwidth. Just possibly they should be grateful we've let them get away with it all these years.

Dual screens, fast updates, no registry cruft and security in mind: Microsoft gives devs the lowdown on Windows 10X

coconuthead

Re: What's in a Name?

TENEX was an operating system, for the old 36 bit PDP-10, developed by BBN and later purchased by DEC. Presumably HP still owns the trademark, so the "X" in "10X" can't be pronounced "ecks" without infringing that.

Built to last: Time to dispose of the disposable, unrepairable brick

coconuthead

Leica R

There are adapters you can get to fit Leica R lenses to Sony FE (their full-frame mirrorless mount, e.g. for the Sony A7 Mk IV). I've seen people use these with excellent results.

Adaptation to a DSLR is harder because the register distance is less than other brands. There used to be a guy who would machine a new bayonet mount for some of the lenses, but I believe he is no longer active due to ill health. Still. quite a few were converted to Canon EF mount.

So really your problem was not that the lenses were unusable, but that you didn't know about these boutique solutions. Had you sold the lenses when Sony still didn't offer many FE lenses, the higher price you would have got would have reflected interest from those who did.

(Please note, I'm not talking about the M rangefinder lenses here. That's a much harder problem, because the steep angle of incidence of the light rays from the rear elements on 35mm and wider M lenses causes problems with the filter stack on top of the sensor. There's a company which modifies Sony cameras to suit these lenses.)

Alan Turing’s OBE medal, PhD cert, other missing items found in super-fan’s Colorado home by agents, says US govt

coconuthead

Re: Inventory

Pulls paperback copy of Hodge's biography off shelf, opens at copyright notice "Andrew Hodges 1983". The book contains extensive discussion of the Bletchley Park work.

Brother, can you spare a dime: Flickr owner sends mass-email begging for subscriptions

coconuthead

Re: Shame

Actually, they *have* been innovating, but not in the service of any sensible business model.

Several years ago they built a machine learning system, with custom hardware (lots of GPUs), to automaticaly add "tags" to images based on the content. We're talking here about something that classified all of the billions of images they must have in a matter of weeks. There were some interesting articles and presentations from the engineers involved. Its application, however, was a PR disaster among the users. A defective training set meant it mislabelled many images, some in an offensive manner. Shoddy UI design and a lack of opt-out meant you couldn't always see what was going on, or manually fix offensive tags on your *own* images. The "autotags" were mixed in with the manually created ones careful users had been using, including some large collections using controlled vocabularies.

Once the outcry started, management refused to back down, because it was seen as a key part of Marissa's strategy of providing automatic upload of phone users' entire "camera roll".

Meanwhile, serious UI bugs went unaddressed. It was clear that the engineering team was working on what they found interesting or would further their careers, rather than what was necessary for the business.

What the site needed all along was sound implementation and cost control.

(It's not clear to me whether Smugmug still owns and runs that farm of classification hardware.)

Das Reboot: Uni forces 38,000 students, staff to queue, show their papers for password reset following 'cyber attack'

coconuthead

insufficient pun quality

The pun in the headline kind of falls flat if you know any significant amount of German. "Boot" is pronounced the same as English "boat", and indeed it is the same word, complete with the ship/boat distinction and the exception for submarines.

Remember the Dutch kid who stuck his finger in a dam to save the village? Here's the IT equivalent

coconuthead

Re: About the same time that ...

The preferred UK and Australian spelling for both meanings is "dyke", with "dike" marked as "also" in all 3 of my dictionaries (Macquarie Australian, Oxford Australian and Concise Oxford UK) for those dialects. My Webster's 3rd ed. lists "dike" as the preferred spelling for the "levee" meaning, with "dyke" as alternate, and "dyke" as the only spelling for the slang for "lesbian". This makes sense if the slang version made its way into US English from the Commonwealth dialects.

A dam is not a dyke, which holds back the sea and not a river.

Internet world despairs as non-profit .org sold for $$$$ to private equity firm, price caps axed

coconuthead

TLD allocation should never have been left private and US-based

It's looking like a historical mistake now not to have placed governance of the Internet under the ITU (International Telecommunications Union). It nearly happened and was a close thing. Some western countries were concerned that it would allow internet censorship in other countries, but that happened anyway.

I've had the same mobile phone number for decades, and, like most people, it's important to me that it not change. I don't get my phone provider or some third party shaking me down for money every year just to keep it. No-one would accept that, and no-one should accept it for domain names either.

Bad news, developers: Apple Mac App Store tells cross-platform Electron apps to get lost

coconuthead

Re: CrAPI

The only obvious ways on modern hardware would be to hide everything behind interprocess communication (with its inefficiency) or put the entire implementation in the kernel (insecure, panic-prone, and also often slower due to frequent kernel mode switching).

Long, long ago there was an architecture called the VAX which had 4 rings: kernel, executive, supervisor and user. The kernel had the, well, kernel. Executive mode had the filesystem: you could only call to nominated entry points via a trap, and if it crashed, it couldn't take the kernel with it. Supervisor mode had the command shell, and finally user mode the actual application.

Early Intel x86 had 4 rings too, but no-one used them so I think Intel relegated the unused 2 to an inefficient stub implementation, and have I think now deprecated or already removed them (while adding negative numbered rings for hypervisors). In any case, modern operating systems don't use them.

Now I hear you say that at least the linker symbols should have somehow been obfuscated or removed. But experience shows that even if you do that someone will figure it out and call it anyway. I recall on the aforementioned VAX/VMS systems there was a very popular third-party utility which "knew" the addresses of certain structures in the file system, and needed to be installed CMEXEC (i.e. define its *own* traps to ring 1). Of course, the operating system vendor got blamed when it broke on an operating system major version update when the magic addresses they were using changed.

I think the Electron people had it coming.

Good guy, Microsoft: Multi-factor auth outage gives cloudy Office, Azure users a surprise three-day weekend

coconuthead

authenticator app does not "receive" codes

TOTP authentication does not work by the site "sending" a code anywhere. The code is synthesised on the device (user's phone) from the current time and an initial value set when the authentication was set up. I don't know whether Microsoft's authenticator app also offers some other mechanism, but I use Google's Authenticator app for my Micorosft 2FA, so they do support standard TOTP.

The linked Microsoft announcement says "Users may not receive authentication requests via phone call, SMS or within their authenticator app." Perhaps they meant something internal to their infrastructure was not receiving requests. Or perhaps they meant to write "replies", as The Register assumed, and it is wrong about TOTP access being disrupted.

The safest place to save your files is somewhere nobody will ever look

coconuthead

kind of makes sense, actually

Of *course* she didn't use "My Documents" - the computer is clearly saying anything in there belongs to it, and not her. It would be rude to even look, and anyway she's not working on "documents" like birth certifcates, but whatever it was.

The story would be a lot more interesting if we learned the user's mental model. For example, she might have thought "Recycling" meant reclaiming wasted space and somehow compacting her files for long-term storage.

MacOS wakes to a bright Catalina sunrise – and broken Adobe apps

coconuthead

Re: Adobe Support

Don't blame Apple for this one. It's been possible to ship "universal binaries" which contain both 32 and 64 bit code since Snow Leopard, ten years ago (which one ran depended on your processor). Apple have been hinting and outright telling developers for years to make the move to 64 bit, and the current version of their compiler suite can't even compile to 32 bit any more.

coconuthead

Re: It is not as if Adobe has not had good notice

What is even more bizarre is that Adobe, in their online customer support forums, swear this does not matter:

https://community.adobe.com/t5/Get-Started/Adobe-Creative-Cloud-on-macOS-10-14-3-installs-32-bit/td-p/10303450

Are Adobe simply wrong, and the installations will fail in 30 days when the licenses cannot refresh? Or are they installing 32-bit cruft, that is never run, on new installations? I see the same 32 bit executables listed by Apple's System Information utility, on a system that has never had the pre-CC Adobe software installed. That's consistent with a poster in the above-linked thread who did a fresh installation two days ago and still sees 32 bit software.

coconuthead

Re: 'Trash' is gone

macOS changed "Trash" to "Bin" for Australian English in Mojave. I assume this was a test for the larger UK market. It also supported Brtish/Commonwealth spelling in the UI, e.g. "Favourites" instead of the US "Favorites".

Happy fifth birthday, Windows Insiders! We'd bake a cake, but it might explode without warning

coconuthead

Where did "ring" come from?

Not having had much at all to do with Windows over the past decade, I haven't kept up with the folklore.

Where does the term "ring" come from? Is it from Dante? If so, where are the other 5?

Right-click opens up terrifying vistas of reality and Windows 95 user's frightful position therein

coconuthead

pet peeve

"How to actually login..."

"Login" is the noun for the user's account, "log in" the verb. It's a thing separable verbs do in Germanic languages. Some other examples: "screwup"/"screw up", "mash up"/"mashup" and"rollover"/"roll over".

Newb admits he ran Satori botnet that turned thousands of hacked devices into a 100Gbps+ DDoS-for-hire cannon

coconuthead

Will the right people be punished fairly?

The linked article suggests that he is diagnosed with Asperger's so severe he is on a disability pension. If so, I would certainly hope to see some compassion from the court, and a sentence oriented towards treatment and rehabilitation.

Meanwhile, was the Canadian ISP with 32,000 pwned routers penalised? It was a 0-day, but the routers were from Huawei, and many would say a 0-day was predictable. I don't know about this ISP, but many ISPs have a habit of preventing or making it very hard for their customers to use better-quality routers than the one they supply, all to keep support costs under control while offering the cheapest possible sign-up fees.

Clutching at its Perl 6, developer community ponders language name with less baggage

coconuthead

Let's help them with some name suggestions

I'll start off:

Lana - "LArry's Next Abomination"

coconuthead

same mistake in the new name

Apparently they can't spell "camellia" any more than they could "pearl".

At least "perl" was an acronym, even if all lower-case in some kind of flower-power-era rebelliousness. (By the way, many English speakers from outside the US would naturally pronounce "perl" to rhyme with "peril" if they didn't already know what it was.)

Perhaps stick with "perl++" as suggested by several commenters in the linked blog entry. That will make the pedigree, such as it is, immediately obvious.

Breaker, breaker. Apple's iOS 12.4 update breaks jailbreak break, un-breaks the break. 10-4

coconuthead

Re: Par for the course for Apple now

Here's another one:

10.13 (and I think some earlier) had a bug where if you had fast user switching enabled, switched to the login screen, slept the computer and then woke it, it would clear the username and password fields about 10 seconds after waking, If you weren't quick about it the login would fail, or even worse leave you typing your password in the plain into the username field (because, of course, TAB wrapped).

This was fixed in 10.14, but has reappeared in one of the point releases.

Google to bury indicator for Extended Validation certs in Chrome because users barely took notice

coconuthead

Re: because users barely took notice

File extensions were a thing in the 1960s, long before Microsoft existed.

Operation Desert Sh!tstorm: Routine test shoots down military's top-secret internets

coconuthead

rows of car batteries baking in the 48° heat

It shouldn't be any worse for a car battery to be under shade in 48° heat than in a car parked in 48° heat, which happens all the time. In fact one strongy suspects car batteries are designed to handle 48° heat and much worse.

Farewell to function keys and swappable SSDs in the new two-port MacBook Pro

coconuthead

e-waste

At least in my country, Apple *does* take Macbooks back for recycling. I just brought mine into the local mall.

However, I don't think it's just the lack of recycling people are objecting to. No recycling can be 100%, and then there's the embodied energy making and transporting it.

coconuthead

Re: Function keys are useless, good riddance

Under Keyboard in the Settings app you'll see an easily-missed Modifier Keys button at the bottom right.

This lets you assign ESC to the CAPS LOCK key, which is where it was on the terminal Bill Joy had when he developed vi (that is, to the left of A on the home row).

Of course, now you've lost CAPS LOCK, which *is* a problem for touch typists, and a pain if you're dealing with, e.g., C macros. On the other hand, it was already a pain for touch typists on many keyboards on the Mac, because macOS seems to require the key to be held down longer than for a normal keypress. (It doesn't seem to be all keyboards.)

So, half a solution but it may suit you.

Don't tell Alice and Bob: Security maven Bruce Schneier is leaving IBM

coconuthead

... "Alice and Bob"

but it was Carol you had to worry about.

Must watch: GE's smart light bulb reset process is a masterpiece... of modern techno-insanity

coconuthead

A reason for no physical reset

The reset sequence is insane, but there's a valid reason why they might use one instead a physical switch.

If you have a room full of these bulbs in inconvenient places (like a high ceiling) it could cost real time and money to pull them all down and do the reset. With a reset sequence, you can reset the whole room just by operating the light switch. Especially in a commercial setting, where high ceilings are common and the staff normally present might not have or be allowed to use a ladder, this is a big difference. (Although I wonder how you then go about pairing 42 light bulbs with the control device again.)

And, although I wouldn't want them in my house, I can see that a restaurant might want to run their lighting at one colour temperature for the lunch service and a lower one for the dinner service.

Of course, if the firmware and app could be trusted not to be buggy a reset would only normally be required on installation.

10 PRINT Memorial in New Hampshire marks the birthplace of BASIC

coconuthead

Re: The sign big-notes BASIC

There was nothing high level about the original Dartmouth BASIC. See the language guide linked on this page:

https://www.dartmouth.edu/basicfifty/basic.html

It had no block constructs, or even symbolic labels for GOTO. The dynamic strings and garbage collection, which were genuine advances over many other languages, came much later. (But Lisp had garbage collection and predated BASIC.)

The introduction in the language manual even mentions Algol as a language students might progress to after using BASIC. It was commonly available at the time and a much higher level language than BASIC. Another contemporary higher level language was Lisp.

Dartmouth BASIC allowed students to use a lower-level, less sophisticated language interactively rather than a higher-level language in batch, using the same amount of resources.

coconuthead

The sign big-notes BASIC

BASIC only became "the standard way people ... learned to program" for those who had no standards.

Its legacy has been to give an erroneous impression of how programming should be done to many people outside the field, including those who get into decision-making positions where they can be responsible for spending large sums of public money on software. Programming should not be an endless quagmire of spaghetti code, punctuated by panicked edits and reruns. What might have been excusable in the 1950s had no business existing by the 1980s.

The sign needed only one word:

"DIM"

Reliable system was so reliable, no one noticed its licence had expired... until it was too late

coconuthead

Re: I generate the licenses..

Tree insertion?

Amid polar vortex... Honeywell gets frosty reception after remote smart thermostat tech freezes up for a week

coconuthead

I can think of two good use cases.

Someone who is often away from home on business and doesn't know ahead of time when they'll be back. Turn the heating on before leaving for the flight home.

Someone who has a holiday home, perhaps in a snow resort. Turn the heating on before getting in the car to drive there.

We did Nazi see this coming... Internet will welcome Earth's newest nation with, sigh, a brand new .SS TLD

coconuthead

EU was in ISO 3166 before ICANN added the TLD

A quick search of the ISO 3166 registry shows that EU was added as an "exceptionally reserved" code in March 1998, at the request of the ISO 4217 maintenance agency, so it could be used in ISIN numbers. Those are the numbers which the finance industry use to identify share holdings.

This is 7 years before ICANN launched the .eu TLD in 2005.

If you're going to write (yet another) anti-ICANN polemic, you should take the trouble to research your points. (I'm not a fan of ICANN, BTW.)

"Exceptionally reserved" codes are the ones in the standard that don't meet the criteria for being countries. For example, some are for islands, requested by the International Postal Union. "UK" is another such code, requested by the UK government.

The internet is going to hell and its creators want your help fixing it

coconuthead

Who cares what those guys think on those subjects?

Cerf and Berners-Lee are engineers: why would anyone assume their opinions on geopolitical subjects like this are worth listening to?

In the case of Berners-Lee, we *know* he could not spell "referrer". Do I think he's read more widely than I have on history, totalitarianism, politics and the social issues driving what's happening online? No, in fact I have no reason to think he's better read even than the average person.

The butterfly defect: MacBook keys wrecked by single grain of sand

coconuthead

Apple alupainium keyboard

I'm currently sporting bandages on both thumbs after typing too hard and long on mine. This isn't the first time. There just isn't enough travel and feel. So I've gone back to the "spare" keyboard (bought for a Windows build that hasn't happened) with mechanical Cherry switches, but which unfortunately also has all the key legends written on the front instead of the top, probably because 15 year old boys think that looks cool and they probably don't have all kinds of financial accounts that look unkindly upon passwords being mistyped and therefore ask "secret" personal questions.

I quite liked the later VT220, but the terminal I used and liked most in the VT100 era was the Teleray 1061, which I believe had Hall effect switches. The day came when it was announced all the terminals would be replaced by ICL ones with horrid rubber dome keyboards and ghastly green displays. I appealed to the boss and kept my two 1061s. More of them failed, but with some horsetrading I kept the last two in the building. Then one morning after yet another repair, there was one, with the sickly green glow of an interloping ICL beside it.

Chrome sends old Macs on permanent Safari: Browser bricks itself

coconuthead

80s sockets isn't enough for a modern app

"There's something called the sockets API. Its remained unchanged pretty much since the 80s..."

Funny, getaddrinfo() and kqueue()/epoll() weren't there in the 1980s. So if you want IPv6 support and to avoid the well-known performance problems with select(), you have to use more modern code.

Assuming you have getaddrinfo(), does the implementation and version you have support IDNA? Do you need to do the punycode conversion in your app or is it done for you inside getaddrinfo()?

Have you ever tried implementing Happy Eyeballs using just getaddrinfo() and BSD sockets? I have, and it's ugly, due in no small part to the lack of async interface in getaddrinfo(). Far easier just to call the complete, transparent, implementation Apple provide... which does not use the BSD API.

How about access to the system certificate store? Not even thought of in the 1980s.

DNS discovery? Sure, they could include a complete DNS asynch implementation in Chrome (and, for all I know, do) but there's one written by Apple just waiting.

They probably implement HTTP using a library but Apple give you that too. Again, not in the BSD API.

BTW Apple do support most of their networking improvements in the (what they regard as legacy) BSD API, but there are a few things only supported in their own implementation. Once you start trying to support iOS as well, there's no contest: you can't turn on the 4G radio using BSD, so although present it's essentially useless.

coconuthead

the software may be insecure

And there you have it—"the software may be insecure". The big browser makers, including Google, Apple and Microsoft are collectively forcing the web in general to abandon insecure practices, particularly bad cipher suites and old versions of SSL. This is happening in the browsers themselves, and also at the OS infrastructure level (certificates, HTTP/2).

These companies are quite open, in technical talks and blogs, that this is their aim.

Harsh as it is, I don't want people running old insecure systems online. It's the same principle as forcing old unsafe vehicles off the road. It is bad luck for those individuals but necessary for the greater good.

If they don't aggressively nuke these installations you will get the situation where you have islands of outdated clients talking to outdated servers and modern clients unable to participate. That was the situation for years with IE6 and then with Flash, and Google, Apple, etc. will be well aware of that pitfall.

coconuthead

Re: Mavericks isn't exactly a spring chicken

It's doing networking, and Apple have been fairly aggressively tinkering with their networking stacks (yes, plural, there are multiple ones). For example, after Mavericks the entire domain name resolution subsystem was replaced. Apple also sunset insecure stuff like WEP, SMBv1 and old SSL versions before almost anyone else (because they can, and some big player has to).

I don't know how much of the networking stack Chrome implements for itself, but it's certainly possible there's some API in Yosemite or later they want to use.

This is before we get to other API changes in the UI etc.

So, yes, a program of Chrome's size and complexity probably *is* "intricately linked" to the OS, kernel and other supplied components.

Page:

SUBSCRIBE TO OUR WEEKLY TECH NEWSLETTER

Biting the hand that feeds IT © 1998–2020