It makes me wonder that despite the fast improvements in model capacity (and the claims) we're still using variations on a 9-year old architecture. How is it that we haven't been able to use LLMs to actually improve that?
My approach for AI-first code review, or really any kind of AI technical opinion, is that if the claim AI made is both important and not obviously true at a glance, it has to prove it to me, and keep trying until I'm convinced or can spot an obvious mistake in the proof.
With reviews, this is usually the case where AI is making a claim that something in the PR will fail because of some assumptions or behaviors in code outside of the PR - e.g. "this change will fail in scenario X, because foo is null in this case, because the SQL query doesn't populate it when bar == quux, and it gets propagated as null through the JSON deserialization (optional field)...", where all the SQL and JSON parsing was not part of the code under review, and "bar == quux" is some weird domain special case.
Stuff like this is both critical, and there's no way for me to judge it without an expensive context switch. So I learn to ask for a more detailed walk-through once, and if that doesn't make me "see" it, I just ask it to reproduce it with tests, and confirm it's a real problem. Reviewing the reproduction is usually enough for me to either "see it" or accept they're probably right and ask the author to recheck it.
(Why not jump straight to "reproduce it" for every finding? Because it still takes time to have AI do the repro. It's cheaper than a deep context switch, but not free.)
Personally I don't find programming with C++ that hard. The downside is it needs a brain warm-up, and this is per project, but once that flywheel is spinning, I find it almost effortless to write code.
I have to go through the same warm-up more or less for any language I work with, so it's not that different than writing Python, Go or Java for me.
I‘m out of the loop: we‘ve had Python, Clojure and possibly something else recently. Is that a series by the same people working through several languages? Is it happenstance? Is it a trend, and every programming language is now scrambling to get their own video documentary?
Working through the pros and cons of the W3 Attribution framework and how it might impact ad measurement. I also focused a bit on AppsFlyer as it was something I was curious to think about how the browser based Attribution could affect or overlap with AppsFlyer's recent released web2app attribution models.
I am not exactly waiting for Linux that will have obligatory ads and will take screenshots of my desktop and send them somewhere. Sorry Bill, but now, I've been through this already, I saw how superior DR DOS goes down because your mom was IBM board member, I had to use Windows 98 Millenium Edition, I was lucky to skip Windows Vista. So, again, no, thanks, never again.
Same with your cloud offering, ridiculous solutions like Azure Service Bus that has pathetic performance, pathetic API and high price.
I personally really don't care that you hate podcasts, or don't have the time to click the link and spend five seconds reading the summary, or are just lashing out because you had a bad day—can you do that somewhere else?
The upper division has and is getting an education always has and always will and the same applies to those with money, with the screw worm fly hitting Texas of recent measles is ok fame and the current administration which is the worst in American history run by imbeciles the can do America appears to be gone and education for most along with it.
When it comes to traditional spycraft, I'm of the opinion that everyone does it and everyone has to.
But the First Amendment is a cultural touchstone for Americans. Unilateral disarmament does work in this context. We can pass a law saying we're not going to fuck with social media and propaganda in other countries, or at least other democracies.
Not the first time. There were calls for NPT treaties etc over the decades. It is irreversible by design. Competition and ownership is the driving force.
Good writeup and solid presentation of wifi timing experiments.
With his typical product-ready development and polished descriptions, I'm glad there are also some unfinished ideas in his drawer. (my imposter syndrome)
Couldn't the need for Zones have been solved with ARP-like probing? I.e. if you don't know on which interface to route a link local address, try pinging the address from each interface, and see which one responds.
I once killed the installation of a work laptop, with such an approach.
I used an external SSD disk to run Linux from it, and one day whatever happened while shutting down the OS, killed the partition on the internal drive.
Thus making me having to request a fresh installation of everything, and losing some stuff in the process, as that killed the cryptographic key for the disk as well.
S&P 500 weights are based the value of shares available on the public market not the market cap. Based on that SpaceX will be nowhere near the top 10.
Do you think their valuations wouldn't fall dramatically if they were willing to float a significant proportion of their shares on the market anytime soon.
If a significant percentage of the market is excluded from the index because they don't meet index inclusion criteria, then then index stops being a useful benchmark.
If you change a benchmark whenever you think it'll be 'wrong', then it becomes a measure of the heuristics you use to predict what'll impact the benchmark rather than a benchmark in its own right.
Dude, I had to lie to my doctor and sign the revers to get this magical cure. There was doctor with adrenaline shot on standby, in case I would collapse. And that stuff was latter recalled for causing hearth attacks (astra zeneca). My relative ended in hospital a few hours after shot.
> coming to the exact same conclusions
The conclusion was that vaccinated person can not spread infection. I still have certificate. Is that still a case, or yet another disinformation?
It's surprising that C++'s development trend continues.
When a game or program is made with C++, it's usually nice because performance is mostly guaranteed. But if someone told me to write C++ myself, I'd cry. There's too much to memorize, and the standards are too varied. When I go to a project site for maintenance and it's a C++ project, I instantly lose energy — because it's just too difficult.
I'd be happy if someone else wrote it, but it's not a language I want to write myself
The trick about documentation is depth, not prose.
You need context and understanding to write documentation "like in the old days". No amount of LLM trickery will free you from that. Once you have that source material, it's easy to re-shape it into an 80's/90's/00's doc format.
Negative example:
I was looking into the German manual of my Canon EOS R5 II, and it is just fluff. Hundreds of pages, full of white space, telling me about features without actually explaining what they mean. Awful automatic translations. Their manuals used to be good (looking at my EOS 6D). But these days: oh boy.
