HNHacker News
TopNewBestAskShowJobs

Taniwha

7,647 karma · joined November 25, 2013

submissionscomments
Taniwha··on A few CPU hardware bugs
This sort of bug, especially in and around pipelines are always hard to find. In chips I've built we've had one guy who built a system that would build random instruction streams to try and trigger as many as we possibly could
Taniwha··on 1 kilobyte is precisely 1000 bytes?
Oh sure, and next you'll say a byte is 10 bits ....
Taniwha··on The TSA's New $45 Fee to Fly Without ID Is Illegal
It's security theatre, someone has to pay the performers
Taniwha··on The Cathedral, the Megachurch, and the Bazaar
Yes but Cathedral and the Bazaar was telling us that the world had changed, gcc was free, linux was a thing etc, mainframes (where compilers cost $10k and you (mostly) couldn't bring your own OS) were being replaced by workstations etc.

Commercial access to Unix source was still many thousands of dollars, the whole SCO debacle was an attempt to stop free OSs from being a thing

Many of us who had grown up from the mainframe era wanted to write compilers, work on OS's etc etc it was a hard thing to do (esp. outside the US) before the late 80s, cheap commodity hardware let a thousand flowers bloom

Taniwha··on The Cathedral, the Megachurch, and the Bazaar
Yeah, it's worth remembering that at the time a compiler cost $10k+, an OS $1000s/year - you couldn't work on OS or compiler work unless you worked for a big hardware company - a whole lot of interesting work was locked away from most programmers
Taniwha··on Dealing with abandonware (2024)
You can go to this effort to get it to make the right magic number .... or just find the place where it gets tested and replace the conditional branch with a unconditional branch (or a no-op)

Having said that I once wrote 68k code that was an executable copyright message, and other code that if it discovered it wasn't running on an authorised machine logged to a known port, sent an email, unlinked the binary, queued "sudo reboot", killed the parent process and then exited (all the authorised machines were mine, running a bespoke kernel)

Taniwha··on Pure Silicon Demo Coding: No CPU, No Memory, Just 4k Gates
TT allows you to pay more and build multi-block designs
Taniwha··on VA Linux: The biggest dotcom IPO
It's worth noting that before the dotcom bubble the rule of thumb was that a startup had to have 5 quarters of profit before going public - the whole dot-com thing of going public on vibes before making any profit was part of why it was a bubble, and also why investors were playing in a whole new sandpit and possibly out of their depth
Taniwha··on Adafruit: Arduino’s Rules Are ‘Incompatible With Open Source’
The latest esp32 chips are RISCV meaning you're not paying a licence for the old ESP CPU (which comes from elsewhere).

There are also 10c RISCV cores available that people are starting to use

I don't think it "saves" it particularly though

Taniwha··on The C++ standard for the F-35 Fighter Jet [video]
yup, same for any real time code, new/malloc/free/delete use hidden mutexes and can cause priority inversion as a result - heisenbugs, that audio/video dropout that happens rarely and you can't quite catch - best to code to avoid them
Taniwha··on AI agents break rules under everyday pressure
Guess what, if you AI agent does insider trading on your behalf you're still going to jail
Taniwha··on Having Fun with K&R C
Unix up until at least System III used the structure thing to do the equivalent to unions, before unions were a thing - for example each disk driver had a data structure that began with a common part and then had a per-device part, generic parts of the kernel used a generic structure that only represented the generic data structure, drivers used their own version.

I think part of this is that early Cs didn't have a full type system, there were limits to the types you could describe (the ones you couldn't were mostly a bit esoteric and not very useful) - from memory in the original PCC types were encoded in the compiler as words rather than trees (pdp11 memory was scarce)

Taniwha··on GCC SC approves inclusion of Algol 68 Front End
ESPOL was (is?) simply a version of the standard Algol compiler that let you do 'system' sorts of things.

The Burroughs large systems architecture didn't really protect you from yourself, system security/integrity depended on only letting code from vetted compilers run (only a compiler could make a code file, and only a privileged person could make a program a compiler) - so the Algol 60 compiler made code that was safe, Espol could make code that wasn't, could do things a normal user couldn't - you kept the espol compiler somewhere safe away from the students ....

(there was a well known hole in this whole thing involving mag tapes)

Taniwha··on GCC SC approves inclusion of Algol 68 Front End
Algol60 had call by name, Algol68 doesn't really, it does have "proceduring" which creates a function to call when you pass an expression to a parameter that's a function pointer that has no parameters, you can use that to sort of do something like call by name but the expense is more obvious
Taniwha··on GCC SC approves inclusion of Algol 68 Front End
C originally had =+ and =- (upto and including Unix V6) - they were ambiguous (a=-b means a= -b? or a = a-b?) and replaced by +=/-=

The original structs were pretty bad too - field names had their own address space and could sort of be used with any pointer which sort of allowed you to make tacky unions) we didn't get a real type system until the late 80s

Taniwha··on GCC SC approves inclusion of Algol 68 Front End
Burroughs used an Algol60 derivative (not '68)
Taniwha··on GCC SC approves inclusion of Algol 68 Front End
I would argue C comes from Algol68 (structs, unions, pointers, a full type system etc, no call by name) rather than Algol60
Taniwha··on PCB Edge USB C Connector Library
+copper+ENIG is going to make that 0.7mm
Taniwha··on PCB Edge USB C Connector Library
They say "use 0.6 mmm"

https://x.com/AnasYMalas/status/1982060711670067350

Taniwha··on Linus Learns Analog Circuits
Linus recommends SMD parts kits that are quite expensive, and relatively bulky, I prefer the books of SMD components, they're more compact and ~$15 each and you can buy refills for ~$5.

For things you use a lot of (1k/10k resistors, 1uF/0.1uF caps etc) buy reels, they're surprisingly cheap <$10 from Digikey (if you visit Shenzhen you can pick them up for $2, I bought a complete set years ago for under $100).

I've largely standardised on 0603 parts for hand assembly, I'm older and have older vision, I need a binocular microscope to work - they're worth the investment if you're doing more than a tiny amount of SMD work.

https://www.aliexpress.com/w/wholesale-0608-capacitor-kit.ht...

Taniwha··on Linus Learns Analog Circuits
Yeah I feel exactly the same way about retirement
Taniwha··on M4.6 Earthquake – 2 km ESE of Berkeley, CA
Thanks - that's it
Taniwha··on M4.6 Earthquake – 2 km ESE of Berkeley, CA
There's an SF short story (Larry Niven I think) about a seismologist who predicts the big one, but the math is not quite right, the story gets out, panic ensues everyone heads for Nevada, he guy is still working in his lab trying to figure why the sign on his equation is coming out negative when all the rest of the US falls into the sea leaving just his part of CA
Taniwha··on Chinese astronauts make rocket fuel and oxygen in space
Ions are atoms ..... tiny particles
Taniwha··on The new geography of stolen goods
Showing a completely legal electronics market in HQB in Shenzhen and claiming it's selling stolen phones is rather unfair - there is a building not far from that market that sells and recycles phones mostly by stripping them for parts and rebuilding them from scratch, but it's not that perfectly legal market that is so much fun to shop at
Taniwha··on Here be dragons: Preventing static damage, latchup, and metastability in the 386
I think the metastability can't destroy a chip thing is not true, you can get a flop into a state where it's oscillating at whatever freq the internal feedback path is (maybe up to GHz) rather than resolving to a stable 1 or 0. That can propagate to adjacent flops resulting in a bunch of flops pulling too much current.

Like anything to do with metastability this is a statistical thing - it can do this, but it's highly unlikely.

I worked on a chip in the mid 90s where we were very careful about our clock crossings, dropped in special high-gain anti-metastability flops, designed logic to reduce synchronised signal frequencies etc etc all the good stuff - we calculated that we'd see a failure (and mostly that would be a pixel burble on the screen) every year or so - at the time Win95 couldn't stay up a week so management decided to ship it

Taniwha··on We'd be better off with 9-bit bytes
Yup, M2 had 81-bit instructions, M1 had 72-bit ones
Taniwha··on We'd be better off with 9-bit bytes
Not really - I worked on a DSP with 9-bit bytes in the 90's (largely because it was focused on MPEG decode for DVDs, new at the time) largely because memory was still very expensive and MPEG2 needed 9-bit frame difference calculations (most people do this as 16-bits these days but back then as I said memory was expensive and you could buy 9-bit parity RAM chips)

It had 512 72-bit registers and was very SIMD/VLIW, was probably the only machine ever with 81-bit instructions

Taniwha··on M8.7 earthquake in Western Pacific, tsunami warning issued
Phone just went off screaming with a warning here in NZ - more a "stay away from the water" warning than a "head for the hills" one
Taniwha··on Losing language features: some stories about disjoint unions
Wirth was on the Algol68 committee - I'm sure he understood how those sorts of unions worked.

He also avoided a lot of the more advanced features of Algol68, he thought it too complex, when he designed Pascal

← PreviousPage 4 of 34Next →