kbolino

40 ms·

511.
by kbolino 2y ago
At best, we are talking past each other. At worst, you are misreading everything I write to play gotcha games. Whatever, I'm glad you were able to figure out exactly the right things to do from a first read of a large and complex docum
512.
by kbolino 2y ago
This is not a discussion about what's possible, it's a discussion about what's best. You can write your own opinion here, and it seems like we're in violent agreement, but that doesn't make our opinion GitHub's
513.
by kbolino 2y ago
Nothing in that document states anything remotely like the antecedent of "that", which was: > don't have anything particularly interesting in the .yml file, just the bare minimum plus some small number of uncomplicated scr
514.
by kbolino 2y ago
This is not about YAML in some general or abstract sense, it is about a YAML-based domain-specific language. If you think this is just about YAML, you are hyper-focused on the wrong detail. In order to use this domain-specific language prop
515.
by kbolino 2y ago
> But GitHub recommends that Where?
516.
by kbolino 2y ago
When it gets realistic, with conditions, variable substitutions, etc., it ends up being 20 steps in a language that isn't shell but is calling shell over and over again, and can't be run outside of CI. Whereas, if you just wrote o
517.
by kbolino 2y ago
I also think categorically dismissing LLMs is a mistake. However, an LLM for automated code generation (the context of the thread as I understand it) is basically a dubious-code-copy-paster on steroids. That was already the wrong way to dev
518.
by kbolino 2y ago
The key part of magic numbers is that they appear early in the file. You shouldn't rely on something that will probably appear at some point because that requires reading the entire file to detect its type. A single 0x00 byte, idea
519.
by kbolino 2y ago
The current YouTube certificate has a lifetime of 84 days. The next one will live that long or even shorter. Certificates are rotated too quickly nowadays for pinning to be viable. They'd be better off pinning their CA than their certi
520.
by kbolino 2y ago
And, if you were using a Windows client talking to a Unix server, you didn't want to get a text file in binary mode, since most programs at the time couldn't handle Unix line endings. This is much better nowadays, to the point tha
521.
by kbolino 2y ago
Up until just a few years ago, Notepad on Windows could not handle Unix-style line endings. It probably makes sense now to adopt the as-is convention, but for a while, it made more sense to convert when checking out, and then to prevent spu
522.
by kbolino 2y ago
The number of 9s that can be assigned to these "bounds" currently is zero. They are not even 90% reliable. And there is no straightforward way to get to 90%, never mind 95%, 99%, etc. The sliding scale of reliability you originall
523.
by kbolino 2y ago
No, LLMs do not have "bounded reliability". All reliability figures for LLMs are based upon empirical observation in specific contexts using artificial benchmarks. As they say in finance, "past performance is not indicative o
524.
by kbolino 2y ago
That may mean that RDS could provide Timescale as an available extension without promoting it directly. However, that is a sufficiently gray legal area that I doubt Amazon wants to wade into it (instead of just making a competing product)
525.
by kbolino 2y ago
There are two problems with this comparison. First, probabilistic prime generation has a mathematically proven lower bound that improves with iteration. There is no comparably robust tuning parameter with an LLM. You can use a different mod
526.
by kbolino 2y ago
Great link! It seems there are 3 key things which distinguish the modern "unified memory architecture" from its predecessors: 1. Pointer passing instead of buffer copying 2. Separate bus for memory access alongside the P
527.
by kbolino 2y ago
But licensing is a very obfuscatory word. What are the terms of the licenses? Consumers don't really know and they vary widely anyway. Buying may have been a misnomer but it had some useful baggage that is shed with the terminology s
528.
by kbolino 2y ago
I believe, but don't know for sure, that classic iGPUs still behaved like discrete PCI devices under the hood, and accessed RAM using DMA over PCI(e), which is slower than the RAM is capable of, and also adds some overhead. Whereas, mo
529.
by kbolino 2y ago
A 4096-bit RSA key is still well beyond the means of even a very capable state actor. The standard nowadays is 2048-bit RSA keys, cracking of which is also (probably) still beyond anyone's capabilities. Maybe a multi-year effort direct
530.
by kbolino 2y ago
The purpose of the control would be to establish whether the competition is actually any better.
531.
by kbolino 2y ago
Some extensions, like TimescaleDB, have licenses explicitly forbidding their use in hosted products which compete with the companies backing them. In concrete terms, this means they can't be used in any of the major cloud providers&#x2
532.
by kbolino 2y ago
While 3:2 pulldown isn't atrocious, a 120Hz screen does have the advantage that it's a simple multiple of 24 fps, which can't be said of 60Hz.
533.
by kbolino 2y ago
I don't think Darwin has been directly distributed in bootable binary format for many years now. And, as far as I know, it has never been made available in that format for Apple silicon.
534.
by kbolino 2y ago
It seems like it's time to revise DMARC so that it requires (or at least can be configured to require) both SPF and DKIM to pass.
535.
by kbolino 2y ago
Terraform providers seem to work pretty well, but as far as I know, they're basically separate executables and the main process communicates with them using sockets.
536.
by kbolino 2y ago
Scores that high mean very little, because a) there are so few people who might qualify for such high scores that they can't be assessed reliably and b) the people who wrote the tests weren't at that level and thus could not write
537.
by kbolino 2y ago
Yeah, this seems the most likely thing to happen here. You'll be forced to disable ADP to continue using iCloud in the UK. This still leaves the question of tourists and other visitors, but it at least fits within the parameters of the
538.
by kbolino 2y ago
Assuming the enclave can receive OTA firmware updates and those updates can completely compromise it, which are not actually proven facts, there's no way to target this to the UK alone without either exempting tourists and creating a b
539.
by kbolino 2y ago
You would need to have a completely trusted software and hardware stack to actually own the keys. And that is already hard enough to get on a PC where ownership still means something, it is not going to happen on most mobile devices. To w
540.
by kbolino 2y ago
BYOK does not provide any additional security over the Secure Enclave (and similar security coprocessors). In fact, unless the Secure Enclave were to directly accept your input and bypass the OS, BYOK is worse because the software can jus

More ›