iiinova — where your digital self keeps evolving

Laurence Tratt @ltratt.bsky.social · 21/09/2026

It was amusing to read that the current "AI makes lectures irrelevant" debate has echos from 1781 as related by Boswell.

083

Laurence Tratt @ltratt.bsky.social · 17/09/2026

I am unconvinced it is possible to learn LaTeX; one merely becomes a participant in its ongoing experiment to see how arbitrary a tool interface can be and still attract users.

050

Laurence Tratt @ltratt.bsky.social · 15/09/2026

On my list of things to learn today was not "for tight loops on modern x64 CPUs, I can observe a ~1.5% win to aligning the loop to the end of a cache line on AMD and a ~0.1% win on Intel aligning to the start of a cache line", but I have.

0141

Laurence Tratt @ltratt.bsky.social · 05/09/2026

An unexpected outcome (to me): it seems that I might be able to work out what things current LLMs don't know about, and perhaps can't even work out, by the LLM-generated "please answer this for me" questions I'm now getting.

040

Laurence Tratt @ltratt.bsky.social · 28/08/2026

For those wondering, yk is chugging away: support is broadening and in the last couple of days I've merged things which speed up interpreters by at least 5% (some by quite a bit more). There is a lot more juice to be squeezed, of course, but every little helps!

031

Laurence Tratt @ltratt.bsky.social · 24/08/2026

I can explain that map more easily than I can explain ticket pricing, that's for sure.

010

Laurence Tratt @ltratt.bsky.social · 23/08/2026

I briefly thought there was one part of this that was right, but I realised I was wrong.

110

Laurence Tratt @ltratt.bsky.social · 23/08/2026

You're not the only person! As a non-pillow person I struggle to empathise, but I suspect this just shows how callous I am.

020

Laurence Tratt @ltratt.bsky.social · 22/08/2026

Great advice! Re pillows: no need for the damned things at all if you ask me.

210

Laurence Tratt @ltratt.bsky.social · 06/08/2026

The video of my QCon talk from this March is now up www.infoq.com/presentation...

041

Laurence Tratt @ltratt.bsky.social · 04/08/2026

One thing that has long bedevilled software is that we have to make choices now knowing that they might be bad in the long run. One way of using LLMs is to look into the future: rapidly prototype a change, imperfect though it may be, to understand what choices might bite later.

050

Laurence Tratt @ltratt.bsky.social · 27/07/2026

It worked 1 time out of 6 for me bsky.app/profile/ltra..., though I suspect it didn't "work" as much as "the check was ignored"... It also didn't work for most, in some cases all, other passengers. In some airports it is also comically badly thought out, almost like they want it to fail...

020

Laurence Tratt @ltratt.bsky.social · 18/07/2026

[Note: I obtained a cheap calibration mic, but I suspect any old mic would have done better than nothing! It doesn't have calibration above 20KHz, so the correction above 20KHz is wrong. My old ears will not notice that.]

000

Laurence Tratt @ltratt.bsky.social · 18/07/2026

DRC is at https:// drc-fir.sourceforge.net/. The manual is, admittedly, somewhat inscrutable, but your favourite LLM will quickly tell you how to make use of it and how to draw graphs like the one above.

110

Laurence Tratt @ltratt.bsky.social · 18/07/2026

A couple of months ago I mangled together a sequence of parametric EQ settings that corrected my (horrible) office acoustics. It turns out that what I really wanted was DRC, which creates an impulse response, which is far more complete! The before/after is quite astonishing.

120

Laurence Tratt @ltratt.bsky.social · 12/07/2026

I meant the tissue one blows one nose in ;) I agree with your main points, though.

020

Laurence Tratt @ltratt.bsky.social · 12/07/2026

Conceivably it could matter in the same way that tissues matter. I need tissues, I need them not to be terrible, but I don't care much beyond that -- tissues are largely interchangeable. [This is all very much "maybe if, in a particular future, given certain contingencies" etc.]

100

Laurence Tratt @ltratt.bsky.social · 11/07/2026

If so -- and, even if it's not possible yet, it is now plausible to imagine it -- then arguably languages would matter little, if at all. So we might find we are liberated from something that we no longer care about being liberated from!

220

Laurence Tratt @ltratt.bsky.social · 11/07/2026

I needed to upgrade my Samsung SSD's firmware. The ISO didn't work when written to a USB disk. Fortunately formatting the disk as MSDOS and copying files from the ISO straight to the USB worked a charm. The firmware upgrade itself took under a second!

000

Laurence Tratt @ltratt.bsky.social · 08/07/2026

Definitely some of that; but also, I think, because you listen more carefully because it feels -- is! -- more novel.

000

Laurence Tratt @ltratt.bsky.social · 08/07/2026

Music can become over familiar; a break of a few years allows a fresh look. Nevermind popped up this morning, and within 4 tracks I heard rhythmic variations (SLTS) and brilliantly bonkers panning (Breed, and not the obvious solo!) I had never previously noticed. A new treat!

100

Laurence Tratt @ltratt.bsky.social · 07/07/2026

They're almost certainly using wpa_supplicant, which is not an easy bit of software. On the plus side, several times I've been able to wrangle wpa_supplicant into working where Windows/Mac don't... [Why this stuff is such a mess is above my pay grade.]

030

Laurence Tratt @ltratt.bsky.social · 30/06/2026

New post: Local Reasoning for Global Properties. Could programming language design come to the rescue? tratt.net/laurie/blog/...

1172

Laurence Tratt @ltratt.bsky.social · 25/06/2026

Just pointed at this HN comment: no, I haven't ever taken over LuaJIT! We did have a project some years ago to look at extending LuaJIT on which Tom Fransham did excellent work. Alas, the funder's priorities moved on (as is their right!), so we didn't get to finish it.

070

Laurence Tratt @ltratt.bsky.social · 24/06/2026

This morning I am noticing 20-30 minute delays on well-known sites _sending_ emails. That makes email login codes problematic when they have a 5-15 minute timeout...

030

Laurence Tratt @ltratt.bsky.social · 18/06/2026

It sounds a bit like the "chore", "feat" things one sometimes sees in commit logs. I still have to read the full subject line (at least) so the prefix never seems to actually help me!

120

Laurence Tratt @ltratt.bsky.social · 18/06/2026

Linux sometimes surprises me.

020

Laurence Tratt @ltratt.bsky.social · 15/06/2026

One of my favourite pages on the internet, to which I've returned often over many years, is www.greenend.org.uk/rjk/tech/pol.... I sent a small comment to its author at the weekend and, within an hour, the page was improved further! We are so lucky to benefit from this community spirit.

040

Laurence Tratt @ltratt.bsky.social · 11/06/2026

That too! The oracle approach depub uses is very literal!

110

Laurence Tratt @ltratt.bsky.social · 11/06/2026

Probably depub should be based on some sort of AST walking approach (ideally with data from rustc), and then it could interpret lints saying "except this one". Technically one could put "except X" into an oracle on a case-by-case basis right now, but that is not a nice hack!

110

Laurence Tratt @ltratt.bsky.social · 11/06/2026

I put crates.io/crates/depub together for this a while back. It's crude, but surprisingly effective.

181

Laurence Tratt @ltratt.bsky.social · 10/06/2026

I'll be speaking at London Systems on Jul 2nd. Attendance is free, and you get a lightning talk or two as a bonus!

040

Laurence Tratt @ltratt.bsky.social · 09/06/2026

New post: Test-case Reducers Are Underappreciated Debugging Tools. The more I've used these tools, the more ways I realise they can be used! This post starts at the basics and works up to things like reducing on properties other than just the length of the input. tratt.net/laurie/blog/...

0242

Laurence Tratt @ltratt.bsky.social · 08/06/2026

In Madrid they appear to be fully manning immigration booths, and at least at exit, they're not bothering to do fingerprints or photos. I suspect this is a deliberate, pragmatic, decision to stop the system gumming up.

010

Laurence Tratt @ltratt.bsky.social · 08/06/2026

The thing that sticks in my mind was at Malpensa. A very friendly lady was desperately babysitting passengers through the process hoping to find someone for whom it worked. She wasn't having much luck.

020

Laurence Tratt @ltratt.bsky.social · 08/06/2026

From my perspective, none of this caused me any problems: the delays I experienced were manageable. But it's a very interesting example of a system being rolled out despite being very clearly broken. I wonder if we'll get an interesting write-up of it at some point?

140

Laurence Tratt @ltratt.bsky.social · 08/06/2026

A confluence of events meant that, unusually, I've travelled three times to the EU in the last month. The new Entry & Exit System has failed 5 times out of 6. Each country/airport implements it differently, some comically incompetently. One, clearly, was ignoring it.

220

Laurence Tratt @ltratt.bsky.social · 06/06/2026

pizauth-1.1.0 out with a number of features: systemd users will want to enable the `systemd` feature. I've also updated my web site with signed ssh signatures, so you can tell that I did (or didn't!) make a given release. tratt.net/laurie/src/p...

040

Laurence Tratt @ltratt.bsky.social · 02/06/2026

Even so, I don't want to pretend it's ever _easy_: you need to find a sequence of bytes that is valid machine code of the kind you want to execute; you need that page to be marked X; and so on. Humans find this hard to reason about IME. Agents...? We'll see...

010

Laurence Tratt @ltratt.bsky.social · 02/06/2026

"Memory safety" is an elastic concept, so it depends on the context. In some situations, if you can knock a program off course, you can then chain other weaknesses to gain control of the PC, at which point all bets are off.

110

Laurence Tratt @ltratt.bsky.social · 02/06/2026

Integer overflow is a harder route than buffer overflows, definitely, but as soon as you've knocked a program off course, you have a chance of completely taking it over (e.g. gadgets etc). I don't disagree about prompt injection, but probably only a relatively small % of programs could be affected.

110

Laurence Tratt @ltratt.bsky.social · 02/06/2026

If we magically fixed all memory safety errors overnight, people would move to the next rung up on the ladder. IMHO that's integer overflows. Same scale as memory safety? Nope... but widespread (including in languages like Rust) and underexploited right now AFAICS.

130

Laurence Tratt @ltratt.bsky.social · 01/06/2026

This weekend I visited The Herschel Museum of Astronomy in Bath: a small but perfectly formed museum in the building where he discovered Uranus. What I did not know is that he discovered infrared; if I had known, I still would not have expected to see the actual device he used!

1272

Laurence Tratt @ltratt.bsky.social · 23/05/2026

I'd be very grateful for further tests now, before I consider a more formal release. If you're an rsync user, please feel free to give it a spin and report back to me! github.com/ltratt/rsyncp/

000

Laurence Tratt @ltratt.bsky.social · 23/05/2026

My simple shell version of "ETA for rsync" found more users than I expected. After some useful suggestions, I've rewritten it in Rust, and it now shows sends (green)/dels (red) as it progresses, doesn't stall on big files, and has a (better) sliding-window ETA.

151

Laurence Tratt @ltratt.bsky.social · 19/05/2026

As a long-term sceptic of debuggers, I am still astonished at how often, when I have a weird problem, two minutes of rr (the reverse debugger) pinpoints the problem.

180

Laurence Tratt @ltratt.bsky.social · 19/05/2026

When free wifi on trains in the UK first arrived it was amazing, mostly, I now realise, because folk didn't know it existed. When they did it became unusable. 4G coverage of the route I take is now so good that tethering my phone is now flawless, but not all routes are so fortunate...

110

Laurence Tratt @ltratt.bsky.social · 15/05/2026

I must admit that I haven't seen this particular outcome, and it seems a bit out of character, at least for the default (+better) models. But there is a high degree of unpredictability, so nothing is certain!

010

Laurence Tratt @ltratt.bsky.social · 15/05/2026

Which tool & model is doing this?

110