Playing with fun tech.
Mastodon: @mmastrac@hachyderm.io (https://hachyderm.io/@mmastrac)
Email me: matthew@mastracci.com
LinkedIn: https://www.linkedin.com/in/matthewmastracci/- I used Miri for some key deno libraries and spent a fair bit of time cleaning up the violations. Many of them were real unsoundness bugs due to reference aliasing.
Unsafe code absolutely needs Miri if the code paths are testable. If not all code is Miri-compatible, it's worth restructuring it so you can Miri test as much as possible.
Note that Miri, Valgrid and the LLVM sanitizers all compliment each other and it's really worth adding all of them to a project if you can.
- I did a huge chunk of work to split deno_core from deno a few years back and TBH I don't blame you from moving to raw rusty_v8. There was a _lot_ of legacy code in deno_core that was challenging to remove because touching a lot of the code would break random downstream tests in deno constantly.
- Officially? You can't get a license.
Unofficially? https://github.com/kholia/OSX-KVM
- I've been trying to do something similar to set up Windows VMs with developer tools. This would be awesome if there was a way to inject a `ps1` script where we could go through the awkwardness of installing choco and various dev tools.
For anyone interested, the magic incantation in the autoattend.xml is:
Redirecting to COM1 is a fun hack I discovered that allows you to remotely monitor these from build scripts.<settings pass="specialize"> <component name="Microsoft-Windows-Deployment" processorArchitecture="amd64" publicKeyToken="31bf3856ad364e35" language="neutral" versionScope="nonSxS"> <RunSynchronous> <RunSynchronousCommand wcm:action="add"> <Order>1</Order> <Path>cmd /c powershell.exe -ExecutionPolicy Bypass -File A:\scripts\setup-dev.ps1 > \\.\COM1</Path> <Description>Run dev setup script</Description> </RunSynchronousCommand> </RunSynchronous> </component> </settings>Even better would be figuring out how to slipstream the choco packages into the ISO - it's not super reliable to install these packages in my recent experience.
- 3 points
- I wrote https://github.com/mmastrac/clitest because I needed a more complex testing harness for CLI tests that does something similar. It's not exactly the same, but it's definitely in the same universe.
One-file-per testcase like `tc` does works, but it tends to fall apart a bit at large scale in my experience.
- In the pre-AI days I worked on a system like this that was constructed by a high-profile consulting team but continuously lost data and failed to meet even the basic standards.
I think I've seen so much rush-shipped slop (before and after) that I'm really anxiously waiting for this bubble to pop.
I have yet to be convinced that AI tooling can provide more than 20% or so speedup for an expert developer working in a modern stack/language.
- 3 points
- This is incorrect. Swift optionally compiles to _bitcode_ (+), which is pre-compiled by _Apple_ before distribution, not by your device itself.
(+) Bitcode is now deprecated: https://digital.ai/catalyst-blog/navigating-apples-bitcode-c...
- I think this is an "everything sucks here" kind of story.
We don't understand the immune system enough to make transplants less risky.
We don't seem to know if QoL is better between those who take the procedure vs those who don't.
The ongoing costs to supporting these operations are crazy and the dysfunctional US system doesn't help.
- sudo is not fully battle tested, even today. You just don't really see the CVEs getting press.
https://www.oligo.security/blog/new-sudo-vulnerabilities-cve...
- Either it would generate a more robust (and likely more recognizable) solution, or it would fail to converge, really.
You may need to train on a smaller number of FPGAs and gradually increase the set. Genetic algorithms have been finicky to get right, and you might find that more devices would massively increase the iteration count
- Publishing private correspondence with single board member(s) is super distasteful because the opinion of one member is not the opinion of the whole board. Sure, he got tacit agreement from one, but that's not agreement with the organization as a whole.
That's putting aside how gross it is for your personal comms to leak in public when you might be a little more candid about what's going on.
How can you trust someone who's willing to violate your privacy like that?
The whole drama is interesting as an outsider, but I can't be left without feeling that newPebble is trying to jump start a commercial venture via shortcuts.
Rebble was never going to change the world but they seemed to be very good at maintaining status quo + many small benefits and just reliably serving that.
https://mmastrac.github.io/blaze/
(the API is undocumented but stupidly simple: an async js_read() function and a sync js_write() function)