rbehrends

7 ms·

1.
by rbehrends 26d ago
I think this oversimplifies the problem of dealing with performance issues. In my experience, there are three aspects that contribute to the software engineering cost of performance optimizations: 1. Identifying the cause of poor performanc
2.
by rbehrends 1mo ago
> This makes perfect sense, but that conflicts with the impression put forward by Anthropic and OpenAI (in particular) that they alone occupy 'frontier model' spots. Not necessarily. Even near the frontier, we don't really
3.
by rbehrends 1mo ago
> I understand the argument of "people are not actively looking", but isn't the cost for such a scan getting lower by the week, and Anthropic's Project Glasswing is supposed to find them quite a while ago? You have to
4.
by rbehrends 1mo ago
The relevant legal provision is from §8 TDDDG [1]: "(1) It is prohibited to possess, manufacture, make available on the market, import, or otherwise bring within the territory to which this Act applies, telecommunications equipment tha
5.
by rbehrends 2mo ago
First of all, yes, labor is taxed much higher in Germany than in the US (PPP adjustment already takes VAT on prices into account, so that doesn't matter in this case). In fact, Germany has one of the highest taxations of labor in the w
6.
by rbehrends 2mo ago
> American workers still get higher salaries than elsewhere in the OECD and growth in such industries isn't out competing those same industries in other OECD countries. I used to think that also, probably because FAANG salaries had
7.
by rbehrends 2mo ago
It means that this is a cabinet decision, not (yet) legislation. It still has to go through the Bundestag. Given the opposition within the SPD and the idea being very unpopular among voters, it is not yet clear whether this will actually be
8.
by rbehrends 2mo ago
I am not sure I'm buying this. The raison d'être for our existing software engineering methods is that humans make mistakes and we needed to contain the effects of these mistakes; and without an appropriate methodology to do that,
9.
by rbehrends 2mo ago
My concern with most of these visual benchmarks, popular as they are, is that they are likely more indicative of knowledge (i.e. how comprehensive the training data is and how well it can be retrieved from the model) than of reasoning abili
10.
by rbehrends 4mo ago
I mean, Western providers such as Fireworks AI/Microsoft Foundry (US) or Tensorix (EU) already are offering many of these models on their own hardware with all the typical compliance boxes ticked through a standard API. Either as open
11.
by rbehrends 4mo ago
> For instance, OP's definition H = Var[G] / Var[P] seems to bypass the issues you mentioned: No, this is exactly the definition I am talking about. The problem is that while theoretically you could work with Var(G)/Var(P)
12.
by rbehrends 4mo ago
> That matches what I assumed it meant, and it seems like OP and the post are arguing that that is some kind of surprising interpretation. The unintuitive part is that in quantitative genetics, heritability is defined in terms of varian
13.
by rbehrends 4mo ago
Heritability has a very specific meaning in quantitative genetics [1], which in many ways is not what your intuition would suggest [2]. It is this usage that the article talks about that. That said, there are plenty of critiques of this d
14.
by rbehrends 6mo ago
No, the problem is that when logging in, the provider's website can provide an authentication shell command that OpenCode will send to the shell sight unseen, even if it is "rm -rf /home". This "feature" is com
15.
by rbehrends 6mo ago
I am more concerned about their, umm, gallant approach to security. Not only that OpenCode is permissive by default in what it is allowed to do, but that it apparently tries to pull its config from the web (provider-based URL) by default [1
16.
by rbehrends 10mo ago
There is the Parliament's legislative train website [1]. However, it only tracks actual legislative steps, not the intra-Council negotiations, so the proposal's page appears to be have been largely inactive since 2024 [2]. [1] ht
17.
by rbehrends 1y ago
> The shell, perl and python are likely for scripting and not used during runtime. Some git subcommands are implemented in these. git filter-branch is a shell script, git cvsimport is a Perl script, and git p4 (perforce interop) is a Pyt
18.
by rbehrends 1y ago
> Of course the member states drafted and agreed to those and that's why pressure should be on governments to stop hand over the keys to Brussels. What specific example are you thinking of where additional power was handed to Brusse
19.
by rbehrends 1y ago
> The EU does not so as time passes the EU's power keeps creeping up. Actually, the EU has the same concept of enumerated powers (called "competences" in the case of the EU). They are listed in articles 2-6 TFEU [1]. You m
20.
by rbehrends 1y ago
What you are proposing would amount replacing the current bicameral legislature (with the European Parliament as the lower house and the Council of the EU as the upper house) with a unicameral legislature. That would actually make it easier
21.
by rbehrends 1y ago
Aside from the often cited nullability issue, here is an (incomplete) list of important things that Kotlin still does better than Java: - First class, fully functional closures. - Non-abstract classes and methods are final by default. - Nam
22.
by rbehrends 2y ago
The same argument would apply to any non-compacting allocator, because the worst case memory blowup due to external fragmentation is huge. But such cases are extremely rarely observed in practice, so people use e.g. standard malloc()/f
23.
by rbehrends 2y ago
What happens under the hood is that a `sequence {}` call creates an instance of `SequenceScope`, which has `yield()` and `yieldAll()` methods. When executing the block, `this` will reference that particular instance and `yield()` is essenti
24.
by rbehrends 2y ago
As somebody who uses and likes both Kotlin and Python (and quite a few other languages), I'd be cautious with using a subjective term such as "more expressive", too, but I can possibly shed some light on where such feelings c
25.
by rbehrends 2y ago
Like this? sequence { for (x in 0..<10) for (y in 0..<10) yield(x*y) }.toList() Now, technically, Kotlin doesn't have list comprehensions, only the equivalent of generator expressions in Python, so you have to tack an ex
26.
by rbehrends 2y ago
> the 'yield' keyword Am I missing something here? $ cat fib.kts fun fib() = sequence { var a = 1; var b = 1 while (true) { yield(a) a = b.also { b += a } } } println(fib().take(10).joinToStri
27.
by rbehrends 2y ago
> Note: I wasn't criticizing the paper. I was criticizing the comment, which claimed "it’s better" to view these as special cases. I didn't assume you were. My note about it being a good paper was just a general &quot
28.
by rbehrends 2y ago
First of all, I recommend giving the paper a read, because I think you're misunderstanding the claim (plus, it is a very good paper). The claim is not that they are equivalent, but that tracing GC and reference counting are dual soluti
29.
by rbehrends 2y ago
You can express immutability simply by making all instance variables private and not providing any methods to modify them.
30.
by rbehrends 2y ago
Delaying deletions can result in significant memory overhead [1]. TANSTAAFL, as they say. [1] https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&d...

More ›