HNHacker News
TopNewBestAskShowJobs

jlouis

2,358 karma · joined June 21, 2008

Hacker. Prefers functional programming.
submissionscomments
jlouis··on Why Discord is switching from Go to Rust (2020)
Brewer colour schemes ftw.
jlouis··on Go is not an easy language
Deleting an element from a slice is a O(n) operation. It better be explicit in the program so the programmer knows just how expensive it is. Either you know what you are doing, and then writing it down is trivial, or you don't know what you are doing, and you are creating an efficiency problem for the future. Other data structures have a delete operation because it is efficient on those.

And fully agree on the GC part. GC doesn't remove the work of resource management. It simplifies it vastly. This can some times lull people into thinking they can ignore resources altogether.

jlouis··on Why SELECT * is bad for SQL performance (2020)
Some columns have large size variance. You are hedging against them growing. It's almost never a problem on fixed size columns.
jlouis··on Spotify Optimized the Largest Dataflow Job Ever for Wrapped 2020
The article isn't easy to read unless you have some knowledge up front about the used technologies and what the problem is.

This definitely drives people to comment on other things.

My gut feeling screams they made a problem themselves in the first place which they then "solved". Similar to a "solution running around looking for a problem" type of deal.

jlouis··on Golang's network stack attempts to parse HTTP headers as UTF-8
The string type is utf8.

You would think an old http header is a string. It's not. It looks like a string though :)

jlouis··on Java on Truffle – Going Fully Metacircular
It's just a different abstraction level.

Taking the argument to the extreme, assembly is a misfit.

jlouis··on Analyzing the performance of Tensorflow training on M1 Mac Mini and Nvidia V100
CPUs often outperform specialized hardware on small models. This is nothing new. You'd need to go to a larger model, and then power consumption curves change too.
jlouis··on Pixar's Render Farm
Probably because CPU times fall within acceptable windows. That would be my guess. You can go faster with FPGA or silicon, but it also has a very high cost, on the order of 10 to 100 as expensive. You can get a lot of hardware for that.
jlouis··on Bye Bye Mongo, Hello Postgres (2018)
MongoDB is really easy to set up and the library API is really good. This provides the affordance in the start of a project where you don't have to think that much about system load from users.

You don't have to think that much about your relational model, so you can postpone all kinds of nasty questions about data integrity and multiplicity early on. This enables a team to move fast and break things.

The switch to something like Postgres comes later in the project where you know your requirements better and you need a system where you have more control over your data. You might even have a person responsible for maintaining your data for you on the payroll.

Personally, I usually just grab Postgres these days and use `json` columns for the documents and loosely fitting data, then drag out things into a more relational model as we go.

jlouis··on FreeBSD src repository has transitioned from Subversion to Git
The biggest hassle with svn is adminstrative overhead. You need to give developers access. Git completely works around this hurdle.
jlouis··on I've been merging microservices back into the monolith
Microservices tend to provide good isolation barriers in systems and those tend to increase stability of highly complex systems.

Though Erlang is a counterexample.

jlouis··on Semantics of C in K Framework
Usually it isn't about execution of programs, but analysis of their properties.

To do so you can use eg model checking, and you require a semantic definition for doing so.

I don't recall floating point stuff being hard to encode. See eg the flocq project.

jlouis··on Is deep learning a new kind of programming?
Most people don't really have the data set to exploit larger ML models.

And the mistake people make is that they don't start with a simple model. They move straight into the heavy ones.

jlouis··on AWS Lambda pricing now per ms
I used an extreme to show the point. At 800 ms the savings are also less than closer to 0.
jlouis··on AWS Lambda pricing now per ms
If it takes 60 seconds, that's 600 steps. 601 steps isn't that much of a saving.

It matters close to 0.

jlouis··on AWS Lambda pricing now per ms
Safety.

No function runs forever.

Also if you want to reboot/repurpose the server, 15 minutes is max wait time.

And finally, you don't want people running long jobs here when you have a solution there.

jlouis··on AWS Lambda pricing now per ms
Necessary change. Now writing stuff in fast languages suddenly matter in cost, changing the landscape of when these solutions might become viable for a chase.
jlouis··on Why is Apple's M1 chip so fast?
X86 isn't fixed width instructions. Depending on the mix you may be able to decode more instructions. And if you target common instructions, you can get a lot of benefit in real world programs.

Arm is different but probably easier to decode. So you can widen the decoder.

jlouis··on The Turing-Completeness Problem
This doesn't work. If none of your tms halt you don't know if they would in a little while.

You can simulate this by making a tm which runs 1 step of each machine at a time. This also hints at the equivalence other people have spoken of.

jlouis··on The Turing-Completeness Problem
You make sure all operations on lists and arrays terminate. For instance by requiring the user to annotate where progress is being made on the bounded array.
jlouis··on Denmark to cull millions of minks over mutated coronavirus
1. Yes. This is serious. Mutation is a big red reset button on the pandemic potentially.

2. No. Minks are kept in small cages with a very high density of biomass. This makes it a perfect incubator. Also ferrets have some similarities with humans in the upper respiratory system, making this more of a concern.

jlouis··on YouTube broke for 40% of the UK population after rolling out 60 FPS videos
Probably a complex explanation with many reasons.

I dont think there is a single reason.

jlouis··on Cursed Elixir
This is fun, because it shows just how pervasive function application is in programming. You can build whole programming languages on just function application (which is what the Lambda calculus is about).
jlouis··on Understanding Memory Fragmentation in Haskell
There is a whole memory allocation system used to combat fragmentation toward the OS level. Erlang also uses a multi-tiered approach, but note that a lot of things are easier because there is far less sharing going on and more things are isolated.
jlouis··on The Zen of Erlang (2016)
Arguably, it already is all over the place in telecommunication. But indeed, Huawei being somewhat limited could push more people to buy Ericsson equipment, where a large part of it uses Erlang.

It is strong in the area because you have the perfect storm of large complex protocols bundled with the need for very high reliability and robustness of systems. Few languages does that particular mix as well as Erlang.

jlouis··on Working Group Last Call: QUIC protocol drafts
More importantly, also look at networks that are bad.
jlouis··on Working Group Last Call: QUIC protocol drafts
UDP is as clean as you can get it. It is more or less free of any overhead. And networks know UDP already. A new IP protocol is far more likely to be rejected in the network.
jlouis··on Intel Discloses Lakefield CPUs Specifications
You do the same with fp. First floating point instruction traps and the process is then flagged as needing fp. This is then used in the context switch to also include fp instructions.

The advantage is you avoid storing fp registers unless you are going to use them.

That flag could easily determine what you can run where.

jlouis··on Formalizing Text Editors in Coq
Proof drives understanding of your code. That is why the resulting code is often small and clear. It pushes the field forward, but not in the direction you might think it does.

You see, the shitloads of working code really works and solves problems. But you might not understand why. The work with proofs gets you closer to an understanding of the underlying why-question.

And I don't think PL-research will ever catch up to the real-world experiments done in code. It took 40 years to embrace garbage collection for typical programs, so I don't have hope anything else will move soon.

jlouis··on Featherweight Go
Well, some journals have demanded proof assistant-level proofs of the work they present. This helps the rigor of the notation a lot since you can't cheat on the proof assistant, and if you confuse it, you are in trouble :)

The problem is that getting the notation right is actually quite hard. People write, to the best of their effort, to make the description succinct and understandable. But the notation is pliable, in the same way much mathematics is written: it isn't as rigorous as we would like.

Writing the same thing in regular prose is much much harder and there are many more ways things can go wrong.

So the state-of-the-art demands formal proof which can be verified by a machine. And it shows in the precision of the notation, which often leaves little to no room for wrong interpretation.

← PreviousPage 5 of 22Next →