@rgilton@fosstodon.org
@rgilton@fosstodon.org avatar

rgilton

@rgilton@fosstodon.org

Electronics and software engineer. Using (F)OSS tooling for everything.

Often found immersed in post-rock and Emacs.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

daviwil, to random
@daviwil@fosstodon.org avatar

All the people disappointed with NixOS right now should consider putting their energy into Guix instead.

I would describe The Guix project as "allergic to corporate entanglement."

rgilton,
@rgilton@fosstodon.org avatar

@daviwil I would be all over guix in preference to nixos, but the one thing holding me back is the lack of systemd 😭

azonenberg, to random
@azonenberg@ioc.exchange avatar

Wishing "AI" was long enough to be unambiguous, rather than occurring as a common substring of other words etc.

Otherwise I'd add an email client rule to send anything mentioning it straight to spam.

rgilton,
@rgilton@fosstodon.org avatar

@azonenberg sounds like you need some ML to classify whether something is about "AI"!

azonenberg, to random
@azonenberg@ioc.exchange avatar

Trigger crossbar board shipped before CNY fab shutdown! Hoping it gets out of China before all of the FedEx staff leave for vacation.

Stencil is already here, parts should be coming in the next day or two and should arrive before the PCB.

This will be a fun design to bring up.

rgilton,
@rgilton@fosstodon.org avatar

@azonenberg What are their prices like?

mcdanlj, to random
@mcdanlj@social.makerforums.info avatar

I've just learned that the Raspberry Pi GPIOs can't run in open drain mode.

I've kind of come to take open drain for granted.

Not sure that 3.3V GPIOs will run my external stepper drivers in the CNC mill I'm resurrecting, so I probably should just use Remora instead, probably with an SKR board. Then I can configure those pins as open drain and not worry about it.

rgilton,
@rgilton@fosstodon.org avatar

@mcdanlj could you turn them into open-drains by putting a diode in series with them? Cathode on the pi side, so it can only sink current. Depends on the detail whether it'd work for your application

azonenberg, to random
@azonenberg@ioc.exchange avatar

Had some downtime today and decided to experiment with some new benchtop cable management techniques. These little clips have an snap-on tie wrap that can expand to hold a ton of wires, and attach to the bench with double sided tape.

Not going to replace Kapton tape for one-off stuff but might be the way to go for routing large bundles of e.g. USB and Ethernet cables along edges of the bench and other high traffic locations.

Overview of bench showing some cables secured with straps and others with Kapton tape

rgilton,
@rgilton@fosstodon.org avatar

@azonenberg I wonder if this is a situation that can be improved using magnets? Having a steel desk (with esd mat on top), and magnetic replacements for the kapton tape might work.

frameworkcomputer, to random
@frameworkcomputer@fosstodon.org avatar

What would you like to see from us in 2024?

rgilton,
@rgilton@fosstodon.org avatar

@frameworkcomputer a "low dpi" screen - external monitors are so challenging when the laptop screen does not match their DPI, and more broadly software still isn't really ready for hiDPI.

ktemkin, to random
@ktemkin@chaos.social avatar

periodic PSA:

development board have blindingly bright LEDS?

slap some kapton atop those suckers; it's wonderful as a attenuator/diffuser

rgilton,
@rgilton@fosstodon.org avatar

@ktemkin yes! I often end up with post-it notes sitting on top of them to protect my eyes.

Recently had a board that somehow managed to bleach the post-it note - not sure if it was light or heat from the LED, but it was surprisingly quick at doing it (maybe a week or two).

azonenberg, to random
@azonenberg@ioc.exchange avatar

Working on troubleshooting the scopehal driver for a particular instrument which I won't name to protect the not-so-innocent.

The SCPI stack is stateful across network connections. Which is to say, if you connect to the scope after a socket disconnected for whatever reason, you might send a command and get a reply that a previous application had requested. Needless to say this makes lots of things break.

In order to recover from this I'm about to write some truly gross code that does a task I do all the time in HDLs but have never had to do in SCPI... until now.

PRBS synchronization.

rgilton,
@rgilton@fosstodon.org avatar

@azonenberg ugh. SCPI is the worst! Even if something does conform to the SCPI spec, there's no way for an instrument to tell you the set of commands it supports! You just have to pray that the documentation makes sense and matches the instrument behaviour 😭

rgilton,
@rgilton@fosstodon.org avatar

@azonenberg What really gets me is that people still seem to be implementing SCPI for new devices 😱

rgilton,
@rgilton@fosstodon.org avatar

@azonenberg Separating the data coming back sounds great.

I'm exploring gRPC as an option at the moment as the transport for the new rockface (https://rockface.io) transport.

I think it deals with all of the problems you've described - and
IMO wins over SCPI because it has a schema, complete with types etc!

rgilton,
@rgilton@fosstodon.org avatar

@azonenberg No need for polling :) gRPC supports streaming data.

It uses HTTP/2 underneath, so it doesn't suffer from head-of-line blocking problems. So there can be multiple requests/streams/whatever-the-correct-term-is running simultaneously.

brianokken, to random
@brianokken@fosstodon.org avatar

pytest-param-scope

https://pypi.org/project/pytest-param-scope/

New pytest plugin to allow setup (and optional teardown) around a set of test cases created with parametrized tests.

rgilton,
@rgilton@fosstodon.org avatar

@brianokken interesting. Can't the same thing be achieved with a scoped fixture?

rgilton,
@rgilton@fosstodon.org avatar

@brianokken Ah that makes sense 👍

nedbat, to python
@nedbat@hachyderm.io avatar

New in 3.12: itertools.batched, which finally implements my top-voted Stack Overflow answer (from 2008!)

https://stackoverflow.com/a/312464/14343

rgilton,
@rgilton@fosstodon.org avatar

@nedbat I suspect that's the SO question that I have visited the most! Great news!

azonenberg, to random
@azonenberg@ioc.exchange avatar

Folks who have done both bare metal (RTOS or just simple event loop/interrupts) and embedded Linux system work: why does everyone seem to think embedded Linux is easier?

Having worked on projects doing both I struggle to see why that's the case. Putting down a big SoC and laying out DDR memory buses. Dealing with u-boot, spending forever compiling kernels and libraries and dependencies and device trees. So much overhead and effort.

I can see it making sense for high end stuff like video processing that needs GPU acceleration, complex GUIs that are eaiser to test in a PC environment using the same GUI toolkit, etc. Things that already need GB of external memory or compute power for the application software itself, so you need that overhead anyway.

But for simpler stuff like a web/ssh management interface on some simple embedded system it's soooo much extra work.

rgilton,
@rgilton@fosstodon.org avatar

@azonenberg @mntmn so you've rolled your own crypto? So many security conf presentations about that...

rgilton,
@rgilton@fosstodon.org avatar

@azonenberg If Linux will fit, I will go for it. Brings so many things once it's running. To name a few: drivers, proper process isolation, filesystems, remote debugging via the net, ability to run things on desktop easily, python + everything from pypi, perf.

rgilton,
@rgilton@fosstodon.org avatar

@azonenberg @mntmn sure, sounds like you're well aware of the risks involved 🙂. Sounds like a lot more effort to me though than just using the off the shelf stuff - but hey, if you're having fun!

rgilton,
@rgilton@fosstodon.org avatar

@azonenberg I agree. There's a trade-off between BOM cost and Linux.

For such a small chip, Linux isn't going to make sense. Zephyr would be my next choice - although I've never looked for an ssh server for that!

J12t, to internet
@J12t@social.coop avatar

About now, ’s app is surpassing the size of the . According to mark zuckerberg, it took 7 hours since opening it up for 10 million sign-ups.

Let that sink in. 7 hours for 10 million sign-ups. It took the fediverse … at least 5 years since the standard, and lots of time before that?

This is significantly faster than even the most aggressive projections. At this speed, will be surpassed within weeks not months.

Will they keep their commitment?

rgilton,
@rgilton@fosstodon.org avatar

@J12t will they keep their activitypub commitment? No, of course not! In the short term it makes them look good to regulators.

In the long term they'll kill it off, probably with some handwavy justification along the lines of "spam is difficult", or "monetisation" etc.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • Leos
  • kavyap
  • cisconetworking
  • thenastyranch
  • magazineikmin
  • hgfsjryuu7
  • DreamBathrooms
  • InstantRegret
  • Youngstown
  • slotface
  • PowerRangers
  • Durango
  • everett
  • rosin
  • anitta
  • vwfavf
  • modclub
  • ethstaker
  • khanakhh
  • tacticalgear
  • ngwrru68w68
  • osvaldo12
  • mdbf
  • tester
  • cubers
  • normalnudes
  • GTA5RPClips
  • provamag3
  • All magazines
    •