@codesections As to whole frameworks that do this, I don't know of any but it seems totally reasonable that there would be place for that. (One can argue that ObjC's autoreleasepool functionality is kind of mimicking this—where you can defer releasing memory of refcounted objects that are dead until some performance critical section has ended—but that's not what people usually mean by GC.)

@codesections HotSpot JVM also has a noop GC (openjdk.java.net/jeps/318) for, likewise, short-lived programs (even though the startup cost is unlikely to make that an attractive first choice.)

Not sure many consider PHP particularly worthy of discussion, but even though it's reference counted by default, the cycle collector can be toggled when needed: php.net/manual/en/features.gc.
Same (in both regards) applies to Python too: docs.python.org/3/library/gc.h

So perhaps it's more common than it may seem at first?

@Seirdy Nevertheless, thank you for the opinion, it's a great point to consider. :) (Also sorry about the split status barrage. Guess Mastodon doesn't work well for long-shortform content.)

@Seirdy given OSD's point 9, which I think comes closest to conflict with SSPL. After writing this, I think I've come to the conclusion that for me it is not, in part due to my overall political beliefs. But then again, I'm not the biggest believer in libre software either, so that's not a point I focused on a lot either. I intended for it to be more about the values involved in both OSS and libre software, and how corp's motivations align with them. \

@Seirdy Which also in part disregards the fact that Google is really not shy about publishing a lot of their inventions, therefore making it not that much of a trade secret.

I feel like the point I was trying to make was more beside the SSPL debate and more about whether "open source" has any subjectively good meaning left to it, given OSI's staunch position about SSPL which I don't think is entirely warranted—though it's still debatable whether SSPL meets that at all, \

@Seirdy Totally a valid argument, and I agree to this bit, yet seems like your argument works more on the small scale of hosting software, where you're fine with a couple of instances. The SSPL is (imho) more about forcing the hyperscalers to share their infra for the benefit of others who might want to compete, though now that I write it out it feels a bit moot because, almost tautologically, the only ones who'll really benefit from this are the megacorps themselves and not the average (= small) user. \

Two more things on RP2040:
• Many people are raving about the datasheet (datasheets.raspberrypi.org/rp2), and I'll join them—it's amazing.
• I wonder whether it could've helped in Tom 7's "Reverse Emulating the NES" (youtube.com/watch?v=ar9WRwCiSr). Now that I'd love to watch again, but this time with a better synchronized GPIO :)

Sorry for the rant. This is my anti-npmization sentiment sticking out again. I should revive that rant feed, huh.

Show thread

ESP-IDF is tempting because it's preloaded with useful stuff, but with that comes a cost in that you're constrained by the environment. It's a fair constraint and it's definitely not Linux, but I posit it'd still be fun to write just the bare minimum of code to suit your needs and not pull in the whole scaffolding.

Show thread

This is the reason I was excited for Precursor (crowdsupply.com/sutajio-kosagi) for a long time; it came with the premise of being hackable down to the lowest levels. Though I can't justify putting the money for it down now—I'm pretty sure it'll just collect dust, and a $500 dustpiece is not really a good investment, in my humble opinion—I still find it really really neat. If you wanted to implement Wi-Fi, you'd have to mostly do it by yourself.

Show thread

Like, even though I haven't really written a line of code yet, I already can tell I'll be mildly irritated by the ESP32 because it comes with its own OS and you can't bore down into the bare metal if you want the Wi-Fi to work (which I do) because the driver is effectively written atop that. I mean, it's totally understandable, you'd need to work a scheduler either way, but it still rubs me the wrong way a bit.

Show thread

Too bad I've already ordered a bunch of ESP32s for a side project. Would love to play with the Pico eventually.

Show thread

Because there weren't enough Raspberry Pi puns to go around, "Raspberry and Chips". Thanks Arduino folk. blog.arduino.cc/2021/01/20/wel

Nonetheless, for those who are still seriously using Github for anything that could run afoul of the law, better get off; the Capital Letter Associations have now learned how to use this in their favour, and I suspect this will only get worse over time.

Show thread

So there's a new DMCA abuse on Github that's top of HN: news.ycombinator.com/item?id=2

Hot take here: Github are in the right to take them down—that's what the DMCA requires them to do so that Github don't lose their safe harbor provisions, regardless of the validity of the claims presented therein. (Here the claim is absolute nonsense; they basically say that "the code is used for copyright abuse" but the code itself is not what enables that.)

This was on the horizon for a while already (I remember rumblings of this last year?) but such it comes to fruition. Not to be all "Google is not not evil anymore" but now they have even more data :^)

Show thread

Fitbit is acquired by Google.

There go the last scraps of Pebble.

Show older
Mastodon for Tech Folks

This Mastodon instance is for people interested in technology. Discussions aren't limited to technology, because tech folks shouldn't be limited to technology either!