Posts by Liam Proven

> Per hour? The damn things kept waking up long before then.

Sir does recognise that the speed does not directly imply duration, yes? ;-)

Re: Progress that kills most of us

> It is arguably wrong that a "scientific and technological advance" is a positive thing when it kills us. When your advance kills you, is it still an advance?

Well, yes, agreed.

I hope my use of block-quoting made my clumsy expository addendum very clearly distinct from Gibson's lucent prose.

The point of the Jackpot scenario is two-fold, and the quote I had to hand -- deadlines and stuff, you know -- only described one part.

One is that there is no single big-bang that everyone could point to and say "it happened _then._" It is gradual and accumulative. Lots of little wars, no WW3. Lots of little pandemics, no big Black Death wiping out 1/3 of humanity via a single pathogen. No entire continent killed in one blow, but all of them being decimated repeatedly until nothing big is left.

(For instance, there is a theory -- as in, a hypothesis but with evidence -- that the dessication of the Sahara, which was wet and fertile and populated with modern humans once -- causes dust storms that carry its dust across the Atlantic and thus fertilise the poor soils of the Amazon basin and so created the Amazon rainforest. Before that it was farmed by the ancestors ofthe ancient Incas, Olmecs, Nazca or whatever. Yin and yang: north Africa becomes desert - result, south America becomes forest.)

But _at the same time_ as this is happening, science and tech leaps forward despite the gradual collapse. Nations and governments and economies fall, slowly enough nobody notices. 80% of the people and 99% of the little remaining wildlife dies.

Result, a future with actual thinking AI, robots, invisibility, flying cars, all the staples -- even a uniquely novel form of time travel, because Gibson is an exceptionally innovative thinker -- but not enough to repair the damage, save the global ecosystems, etc.

The points being:

1. Spread the big collapse out over a generation or 2 and people simply won't notice. I find this horribly plausible.

2. At the same time as a global collapse and replacement of democracy and economy, done by stealth with nobody really noticing until it's too late and it's over, R&D continues and keeps making multiple leaps. The point of the 1st Jackpot novel is time travel; the point of the 2nd is true autonomous AI but nobody notices because there was so much fake stuff around.

> I hope you are not suggesting that Liam's buried alter-ego is starting to rise to the surface

Well _now_ I'm worried.

You could have just saved us all time and said "I don't read books", you know.

Re: OCCAM and Transputers

> It's OK, Liam. You're not. I discovered Unix in mid career.

Thank you. :-)

I am getting on a bit, but then again, I have a wonderful 6YO kid and a still new and happy marriage. In some ways, I still feel important parts of life are beginning and unfolding.

Which does mean that despite a nominal retirement age looming, I will keep writing for as long as people find some value in my work -- to keep paying for that kid.

(Until WW3 and total climate collapse, anyway.)

Re: OCCAM and Transputers

> the 'Revised Report on Algol68' by vanWijngaarden et al is probably the single most unreadable and difficult to understand document I have ever had the misfortune to encounter

Van Wijngaarden, the man who killed ALGOL.

He was the Thomas Midgely Jr of computing.

Midgely invented leaded petrol. He stopped engines knocking but poisoned _tens of billions_ of people and countless megatons of wildlife.

Then, if this impact on the world were not enough, as a second act, he invented CFCs and created the hole in the ozone layer.

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

Adriaan van Wijngaarden killed Algol. A language that was loved across the industry and used for everything from OSes to end-user apps. Hoare and Wirth's proposal would have increased its type safety while making it more readable.

AVW's bureaucratese-stuffed version killed it.

That led to the invention of C, BASIC, Python, Java, Javascript, and the pile of poo that is 21st century software.

The result was "lighter" and "simpler" languages, and OSes built from them such as Unix, which resemble the Eiffel Tower built from matchsticks.

https://www.loper-os.org/?p=69

They are amazing and beautiful and so absurdly fragile and impractical it's amazing they remain upright at all. C is so compressed and terse and unreadable, as is the command line of the OS built in it, that it took 2 geniuses to create it.

https://xkcd.com/2347/

C is so famously hard to read that those who learn to are very proud of it.

Result, a burning love for it. It makes geeks feel macho.

https://x.com/smdiehl/status/855827759872045056

None of the thousands who have seize on it since are such minds and they've built a Meccano-and-Lego nightmare from it.

The mainstream software industry is really bad at software.

https://xkcd.com/2030/

And it's all anyone under 50 knows so they don't know how bad and wrong it is.

https://xkcd.com/463/

Wirth was so damned good that he took ALGOL W and crafter Pascal from the wreckage, and it was a hit in its way. Then Modula-2, which is was another, smaller one. Then Oberon, the most amazing overlooked jewel in 1990s software.

https://web.archive.org/web/20231201144315/https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.90.7173&rep=rep1&type=pdf

Oberon is (one of the the things that) inspired Plan 9 and Inferno, the biggest missed opportunities in the entire shaky cityscape of C.

I blame Adriaan van Wijngaarden.

His response? To introduce Wirth on stage with the most famous joke about his name, a joke Wirth did not like, but is now attributed to him:

«

Europeans pronounce his name correctly (Ni-klows Virt), Americans invariably mangle it into (Nick-les Worth). Which is to say that Europeans call him by name, but Americans call him by value.

»

Re: OCCAM and Transputers

> Quite surprised you didn't mention his involvement with INMOS and the Transputer.

I did consider it, but with great regret decided that it was probably too obscure. What I wanted to convey was different, separate elements of his work affected millions, in fact arguably billions, of people, and keep the piece short enough that it touched on some of the notable points of his life and career without being so long that people would skip it. He was quite close to 100 years old, and much of his most important work was before the bulk of people in the industry were born.

I've been interviewed on a few podcasts now, and one of them was LinuxLads:

https://linuxlads.com/episodes/126/

They asked when I discovered Unix and Linux, as if they were the same thing. I said they were two different times: I discovered Unix in the 1980s, at university, and by 1989 was installing and supporting SCO Xenix systems in production -- before Torvalds started work on his toy kernel project.

There was a stunned silence and then a mutter of "oh my god." This hadn't occurred to them as a possibility. I think they'd never met someone who was using any Unix before Linux existed before. It made me feel very old, but I'm used to that these days...

Re: Hoare logic

> He was also known for introducing what is now called "Hoare logic"

I did specifically mention that. I got a message from a friend of mine, thanking me for the obit, because he never realised that the same person devised Hoare logic _and_ Quicksort.

> I have always viewed this as a joke

This stuff is _way_ above my pay grade ,but no, I do not think so. I think he meant that rather than having a value that means "there is no value", it should have thrown an error.

> All I remember is it heavily featured the dining philosophers problem which we may or may not have proved a solution to...

I did consider a mention of the Dining Philosophers, as an example I actually managed to more or less understand somewhat.

Dijkstra formulated it, but Hoare took it mainstream, I believe.

I think someone, possibly not Hoare, once said "Edsger Dijkstra is Dutch, and behaves as such."

Alan Kay said "arrogance in computer science is measured in nano-Dijkstras."

Ow...

Re: is zram backing compressed?

> have been unable to find out if data remains compressed when zram moves data to the disk swap.

I thought it did, but during my research for this article it looks like no, it doesn't. I _think_ it's expanded again so that the existing unmodified read/write mechanisms still work.

For a previous recent piece, I found that the RAM image dumpted into the swap volume for hibernation _is_ compressed. So there is a different code path that can read/write compressed data.

Some unification work here is needed, I reckon.

> Word has one file open - 750MB.

I can only tell you that Word 97 still works, you can activate it with the magic number 1111-1111111, and its file format remained constant for the next decade so everything can open it. I think I reviewed it when it came out and mocked its size, but now, it can run inside any vaguely recent processor's L2 cache.

Runs a treat in WINE, including installing all the Service Releases, and is blazingly fast on an old low-end Core i5.

> Oh well to have a with the DR I guess.

... Pardon?

> Just dumping a copy of the GPLv3 into the root doesn't legally mean anything.

Then go file an issue.

I dimly remember seeing the mailing list post, where the author found that the company no longer existed and rights had reverted to him and he responded by putting it on Github. I think _he_ thinks uploading it and putting the GPL in there counts. If it does not, go tell him, not us. Explain what he needs to do.

Re: DR DOS 3.41 was from 1989, not 1981

> DR DOS 3.41 was from 1989, not 1981

Whoops -- I knew that! A plain old typo. I will notify the editors.

> those are mutually exclusive things.

Agreed.

He talks a lot on Reddit about re-implementing the distinctive stuff that made DR-DOS _DR_ not MS-DOS, but honestly, DR-DOS was a relatively quick hack to have an entry in the DOS market. It was a low-end product, simple and single-tasking, while DR worked on the real stuff, the multitasking Concurrent DOS and real-time FlexOS.

The first version incorporated the big FAT16 volumes from Compaq DOS 3.31 which MS-DOS/PC DOS 3.3 couldn't do.

It sold. It made money. So they leapfrogged the lardy MS-DOS 4 and did DR-DOS 5, which integrated Quarterdeck QEMM386-style upper memory block handling. This was a DOS that could live in the High Memory Area (from 1024 kB to 1088 kB in RAM) and load its own drivers into UMBs. Result, a DOS that _easily_ gave you 635+ kB of base memory, just as DOS apps were getting huge and really struggling to fit into 640 kB along with a network stack, multimedia and mouse support and stuff.

And in a stroke of genius they sold it at retail. You couldn't buy DOS before that. It only sold to OEMs for inclusion with new PCs.

And it had the ViewMax GUI, a cut-down GEM.

DR-DOS 5 sold great. MS noticed and did MS-DOS 5, a copy-cat release.

So DR did DR-DOS 6, with bundled disk compression.

MS copied with MS-DOS 6. We know how that went... 6.2, 6.21, 6.22...

Novell bought DR and one-upped MS again: Novell DOS 7 had built-in peer-to-peer networking.

But now, the main thing that was distinctive about DR-DOS was colon column separators in `dir /w`.

There's not much for him to include.

> It's sort of LaTeX

*starts to whimper*

> on steroids

Oh gods, please, no...

> with a built-in scripting language

*Screams and faints*

Re: extend ...

> I really, really think the encrypted binary blob feature should be added before the usual suspect gets in there first.

I saw the icon, but please, don't give anyone more ideas...

This is a proper "at least it's not raining!" level comment. FFS, _don't say that..._

*Fading scream*

Re: no asciidoc?

> no asciidoc?

You will note, I hope, that I did mention it.

The thing is: Markdown is relatively simple and thus relatively simple to render, and there's lots of code that does it. I am sure _some_ of those renderers are cross-platform and can be compiled.

I have worked with ADoc. I liked it. I came to it from editing DocBook XML directly and the relief was extraordinary: DocBook is like sticking bamboo needles under _your own_ fingernails.

But there are only 2 (?) implementations of ADoc, and it's complicated, because ADoc is designed to render to a narrow subset of DocBook, and DocBook isn't just a complexity hairball vomited up by the bureacrats of the SGML committee... it's one the size of a minor planet.

It may come, in time. I would not hold your breath. RST might come sooner.

The problem and at the same time the benefit is that Markdown is so simple, you can't do much. Whereas ADoc and its kind are designed to be capable enough for a whole book, and that means importing an entire container-ship freighter load of complexity and pain.

> Marketer 1 wants the font size increased by 0.5 pt and moved 3mm left.

> Marketer 2 wants the font size decreased by 0.5 pt and moved 3mm right.

"That's the great thing -- you can't!"

This can be interpreted as a winning attribute of a format with no positioning control and no colour formatting.

It's like TeX but for mortals.

All the formatting you _need,_ and nothing else.

(I know, but it is here and it's real and it sort of works.)

> I don't know if this might help? I've had it running on Windows 10 a couple of years ago.

Quite so. I specifically mentioned it in an article a few years ago LINKED IN THIS ONE!

https://www.theregister.com/2023/08/23/libreoffice_76_is_out/

SmartSuite is officially discontinued. TBH, in my book, that means it is fair game and have at it. I tested it for a reader about 3-4 years ago who emailed me at El Reg asking for help reading Lotuis 10-203 files on Windows 11.

I told them about Smartsuite, where to get it, where to get the Fixpacks, and I tested it for them and verified that it worked.

No reply. :-(

> wrong forum?

I _did_ mention my PiHole in the article.

So far nothing has tempted me away.

It works, it does the job, it's got a nice friendly web UI I barely need to use, and it's 100% FOSS.

So far, the alternatives all look to be more work, or not FOSS, or have poorer UI, or a combination of 2 or more of those.

> I just meant Apple devices.

Aha! OK then.

TBH my rediscovery of RSS is new enough that sync has not yet appeared on my radar.

However, I _badly_ want Thunderbird Sync. This is in development but what for me makes it _especially_ annoying is that Mozilla has the infra, Firefox syncs, Zen and Waterfox etc. talk to it, but T'bird still does not and there have not even been any updates in a year or more.

I don't need _message_ sync. That is the servers' job. Sync of rules and filters would be great but is not essential. Just accounts and server settings would do!

Re: How to find the Register's RSS feeds?

*Mutters resentfully*

That's the display department.

> across my Ecosystem machines

What does this mean in context? Why do you capitalise it?

Re: How shitty can it get? - have a look at physorg

> Physorg, was once a favoured site

I must confess it is not a regular haunt of mine, but I did not know it had got so bad.

The HQ is about 900 metres from me right now -- it is in Douglas in the Isle of Man.

Re: Well Done and Humourous

> Loius Rossman's YouTube channel

Nice. I watch as little Youtube as possible and only follow 1 channel, a Canadian motorbike one. I miss motorbikes.

But yeah, I saw the video first, and only read the report later...

Re: Thank you Trump

> If your articles were influencing official Chinese policy, I think you'd be able to ask your bosses for a big raise!

Or ask Uncle Xi for one.

New job title: Ultra-influencer.

I need a Tiktok, stat!

Re: Thank you Trump

> Your unhinged, rambling and totally unprofessional behavior

I didn't read the title of the post at first, and I thought you meant *me.*

I was wondering what I'd done to affect the entire People's Republic of China. Hell, even the 22M people in the Republic of China, AKA Taiwan... :-o

Re: Bloat - the modern curse

> Taking a casual look at the ‘dd’ utility on the Linux box in front of me, I see that it’s about 72k in size. So etcher is only 5,660 times bigger to do essentially the same job. Way to go!

Exactly.

And the real thing is this. Yes, Etcher slaps a pretty GUI on it.

But so does Rufus -- https://rufus.ie/en/ -- and it's under 2 MB.

Rufus is Windows-only though.

By way of comparison, there's USBimager -- https://gitlab.com/bztsrc/usbimager/

That's got a friendly GUI, _and_ it is cross-platform, _and_ it is FOSS... and it's 300 kB.

Today I read this:

https://www.dbreunig.com/2026/02/21/why-is-claude-an-electron-app.html

It says "Electron has won".

This interpretation of the same piece:

https://tonsky.me/blog/fall-of-native/

Says "we've lost native."

No we smegging well have not. It's laziness and it's lack of skill and it's deteriorating OS UI.

https://take.surf/2026/03/01/welcome-back-to-macintosh

> Another damned fork

Yes.

> this is why open source has such a hard time getting ahead.

No, I don't agree. Sometimes they really add value. MATE is a fork of GNOME 2, Trinity a fork of KDE 3.5, Cinnamon a fork of GNOME 3, Firefox a fork of Netscape, Blink a fork of Webkit which was a fork of KHTML...

All things some folks love.

> Zed actually added an option in their settings a little while back that simply turns off all the AI stuff.

I am aware; I used it myself on my own copy. Gram does not turn it off, it _removes_ it. And that chat, and the Ts&Cs, and more. That sounds good to me.

> Unlike MS, they handled that properly.

Yes, agreed on that bit. :-)

> It then failed to do that

From its own website:

«

There are no automatic updates or downloads.

Extensions must be built from source and do not auto-update.

»

https://gram.liten.app/posts/first-release/

It's the first release, version one point zero, and I guess he has not reworded all the dialog boxes and things, but it sounds like it is doing what it says on the tin: no automatic download of anything.

Re: An alpine-based version of CB++ sounds very interesting

> play nicely with Ventoy?

They have done in the past. I am thinking of replacing CB++ 13 (installed via Ventoy) with BL so I can see shortly.

At the end of the day, it's just Debian. Should be fine.

Re: An alpine-based version of CB++ sounds very interesting

> An alpine based version of CB++ also sounds interesting in terms of being lightweight....

Absolutely.

I suggested Alpine in the CB++ subreddit. Ben said he'd love to do that, but the Calamares installer he uses doesn't run on Alpine. Then he checked and found it was removed from the latest Alpine.

I enquired about that on the Alpine mailing list. There'd been some build problem, but they fixed it and put it back.

I told Ben, and I've not heard back since. I offered to introduce him to the Calamares author, but nothing yet...

Re: The Original CB

> I'm (still) waiting for a modern version with those features.

BunsenLabs has a live session.

I suggested in both distros' subreddits that one or the other did a Devuan-based version, for greater differentiation.

BL did not respond. CB++'s Computermouth said that he was perfectly happy with Debian and saw no need for a Devuan version, since he'd never even tried Devuan.

:shrug:

Re: They're just saving time. Making this stuff quicker and easier.

> It's not very linuxy, either. Do one thing well? Window management is a desktop thing.

1. That ship sailed so long ago you can't even see it over the horizon from a mountaintop now.

2. Most people and most distros don't have tiling WMs and I suspect don't know how to use the tiling features in their OS.

Just 2 web pages, and only 2, side-by-side, is probably all the tiling that 80% of users will ever need. And there you go: Dr Pareto is thus satisfied.

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