As a Yocto enthusiast, I am curious as to how much elapsed realtime would be needed for a clean Yocto build. Yocto is thread heavy, so with 288, it oughta be good.
My Yocto build times on a 32-core AMD are negligible, <2 minutes for a full distro, IIRC. I suspect higher core counts have diminishing returns, especially since most dev builds are heavily cached.
I'd like to believe that Silicon Valley mgmt is Pro-Trump in the same way that Oskar Schindler was "pro Nazi".
You may not personally like who is in office, but you pretend to in order to survive.
This isn’t the case, sadly. Some people, like Ben Horowitz sadly, have gone completely off the deep end.
Some are culture warriors who feel they have been wronged, some are opportunists. But the thing with opportunism is that this is who they are and what they believe in. Having a president who is corrupt is exactly what they want because they know exactly how to work with him: quid pro quo.
There is no distance between them being pro-Trump and opportunistic. He’s the perfect embodiment of those values.
There are a few people like that (we know who they are) but either tech has changed or I never noticed but a significant portion of the senior leadership in the tech world is MAGA (not in the dumb way - but in a far more problematic "techno-libertarian" way)
All the real world libertarianism I've seen trends towards the Elon Musk kind. I almost intentionally use the word since I don't see any other outcome so it's important to use the word liberally :-)
2 things to think about here:
1. Coding is just one phase of the software development life cycle (SDLC). We still have to gather requirements, design, test, release, and most importantly maintain. I was taught, albeit years ago, that code spends most of its life in maintenance and that is the phase where the most money is spent.
2. Keep in mind Amdahl's law. The limit of software cost as coding cost approaches zero is the cost of the other phases of the SDLC. Apologies to Amdahl for the cheap, dirty bastardization.
When I began work at my last company, we all had to badge in to get in the parking lot, where there was 6 lanes 6-10a and just 2 other times. We also had to pass through 1 of 4 turnstyles, and we were subject to bag inspection going both in and out. We were trained to NEVER leave my badge in an unguarded location (ie my locked car at home, at lunch, etc). We were also trained to NOT display our badges off campus, especially when travelling.
This made me make DAMN SURE I knew where my badge was at at times. Same hook in the closet. When walking OUT through the turnstyle, I usually either put it in my bag if I had one, tucked into my shirt pocket, or just tucked under my shirt.
"The problem? I don’t own any RISC-V hardware."
Seems like a YOU problem. There is PLENTY of affordable (< 200 USD) RISC-V Orange Pi RV2 boards available for immediate shipment. http://www.orangepi.org/html/hardWare/computerAndMicrocontro...
Shoot, if you are testing LOTS of architectures, then get a FPGA dev board and test away. I can't testify about the availability of a IBM 360 open source firmware design, but I know for a fact that Xilinx makes RISC-V available.
> Here’s the fun bit: as of right now, there are essentially zero RVA23-compliant RISC-V boards you can actually buy. They’re coming – a handful of promising-looking boards from Sipeed, Milk-V, Banana Pi and others are expected around April/May 2026 – but they’re not on shelves yet. Even the Orange Pi RV2, released in March 2025, is RVA20 only, which means it already can’t run Ubuntu 25.10. Ouch.
I'm not familiar enough with Ubuntu or Snaps to say how much of a pain it would be to just use an older OS version, but if someone is in the Ubuntu ecosystem I certainly wouldn't blame them for not wanting to spend money on hardware that the OS has already deprecated.
(And IMHO the complete lack of available hardware means that Ubuntu made a poor choice here, but that's rather beside the point; this author is working inside a specific existing ecosystem, and their choices seem reasonable in that context.)
Although, thinking further.... I do wonder if it wouldn't be better to explicitly use RVA20 with a compatible OS (probably Ubuntu 24.04) in order to ship binaries that are as compatible as possible. Like, if you ship a snap that uses RVA23 instructions, your app won't work on existing hardware, right? (Whether that hardware is running Ubuntu 20.04, or Debian, or Gentoo, or...)
Makes sense for all the people running Ubuntu under emulated RISC-V who want/need to use snaps for package management, I guess?
Since there aren't boards available that can run 25.10, why even bother packaging for that architecture? Why dig trenches for cable that can't be laid?
I'm glad to see there is a "Beyond the Code" section that discusses comments. Here's what I typically told my students in Intro to Programming"
Good comments lend insight into the code. Reading the code itself tells you the what. Comments should explain the why. Comments like "i+=1; /* Increment i */" are of little value. However comments such as "We increment i mid loop so that we can peek ahead at the next value for a possible swap" are more useful.
Use a narrative voice when writing comments, like you are explaining the code to your grandparent. This make digestion easier.
Remember, code spends most of its life, and most of its expense, in the maintenance phase. The easier you make your code to understand, the less it will cost and the longer it will live.
My first team figured that out after a year or so. If it’s really TODO, it should either be addressed before the WIP feature is considered “completed”, or it needs to show up in our work tracking system. Otherwise it just fell through the cracks and would never be prioritized.
The number of people I know who honor TODOs is pretty slim, but I find it does in fact help if you get people to stick their initials next to it. Over enough refactors a TODO can lose its meaning and author.
TODO (DKH) - This should handle negative numbers
can be helpful later if there's nothing in the requirements about negative numbers. Why do we need to handle negative numbers? Is it for some mystery feature? Performance? Security? What? Why?
'DKH' might recall what they were thinking if you prompt them.
We see what we want to see, and a TODO when we are in a hurry probably won't trigger much guilt or associated motivation to do something about it. Or at least, not as much as if we see our own initials in the TODO.
(Also handy when prepping a commit, since you might have left breadcrumbs in case you get interrupted, and then forget one or two of the tasks you still had outstanding afterward)
I am going to use a term that is hard to measure: authenticity. A lot of TV programming today, not just for kids, but in general, seems to just pander to the audience rather than tell it as it is, with genuine struggle and strife. There are exceptions (The Bear, most notably). But a lot of it just seems like pablum and slop generated to attract eyeballs rather than to tell a story. Kids sense, and dislike, condescension.
Old skool "Sesame Street" was engineered to attract eyeballs, too. They had kids watch an episode and if their attention wandered, they changed the episode. It wasn't pablum, though, so I don't think one implies the other.
reply