Rochus

15 ms·

151.
by Rochus 4mo ago
> in Standard Pascal it was pretty much impossible to write a library implementing linear algebra algorithms You likely mean Wirth's original Pascal; "Standard Pascal" (i.e. ISO 7185) had support for actual array parameter
152.
by Rochus 4mo ago
The Olimex board has 768KB RAM and 32MB PSRAM, more than enough. The original Ceres only had 2 MB of DRAM and 256 KB of VRAM.
153.
by Rochus 4mo ago
> any plans to update for the 5? It currently works on the 2b, 3b and Zero 2; the latter was my actual target because it is a very nice and still lean board available for ~15$ worldwide at least until 2030, and meets very well with the O
154.
by Rochus 4mo ago
> nowadays it places an unneeded toll on the shift and caps lock keys and makes coding a bit more tedious Right. But there are evolutions of Oberon without these orthodoxies (e.g. https://github.com/rochus-keller/o
155.
by Rochus 4mo ago
> Do you think it is good policy for a country to cap its population? That's a misconception. The goal of the initiative is to limit immigration and infrastructure strain. As a Swiss myself I can only talk about my country, and yes,
156.
by Rochus 4mo ago
Paywalled (and apparently globalist lecturing from a glass house).
157.
by Rochus 4mo ago
Your analogy doesn't quite hold. The primary use case of a compilation target is to compile programs, not the compiler itself. With Dart specifically, "dart compile wasm" already exists precisely for that purpose. Compiling t
158.
by Rochus 4mo ago
Why not Asciidoc? It works well with AI so far. To make it even work better I made a more formal subset which can be read with a single-pass parser ( https://github.com/rochus-keller/leandoc/ ), to reduce the "
159.
by Rochus 4mo ago
> I was able to compile the Dart VM (the compiler + analyzer + VM) to wasm and run it on the web Is this really a representative use-case of WASM/WASI? Would'n it be much better to compile Dart to WASM (the Dart SDK even suppor
160.
by Rochus 4mo ago
According to cloc, it's about 15kSLOC of C (body + header), plus 228 SLOC of assembly. xv6 x86 for comparison has about 7.5kSLOC of C and 201 SLOC of assembly. According to your readme, you didn't use coding agents, but just promp
161.
by Rochus 4mo ago
Very interesing; looks like an incredible research effort.
162.
Show HN: A parser for the ISO 10303 EXPRESS language for its 40th anniversary(github.com)
7 points by Rochus 4mo ago|0 comments
163.
by Rochus 4mo ago
Not to be mixed-up with the Micron programming language (see https://github.com/micron-language/specification and https://github.com/rochus-keller/micron/ ), also mostly written in C++. That
164.
by Rochus 4mo ago
Interesting interview, thanks for sharing. It's fascinating that Odersky started with Modula-2 and did a PhD with Wirth, but eventually switched to functional programming.
165.
by Rochus 4mo ago
Interesting documentary, just watching it. Includes a lot of stuff I didn't know and shows a lot of involved people. Great.
166.
by Rochus 4mo ago
Yes, I think the reaction is disproportionate. The philosophical concern is understandable, but withdrawing the project doesn't address it; it just removes something valuable from people who aren't responsible for the problem he p
167.
by Rochus 4mo ago
It's his code and of course he can decide what to do with it. I just think it's a shame when people make hasty, ill-considered decisions based on misunderstandings and false assumptions. Copyright law is (unfortunately) much more
168.
by Rochus 4mo ago
> designed to maximize source availability Right. It depends on what you mean by "use"; GPL maximizes use in the sense that it prevents anyone from taking the code proprietary and thereby restricting future users' access.
169.
by Rochus 4mo ago
> You'll have to prove that LLMs do not actually contain encoded copies In law, the presumption is that an act is lawful unless proven otherwise. The burden lies on whoever claims a violation occurred. I already went into the case o
170.
by Rochus 4mo ago
> LLMs trained on GPL code can produce derivative works This confuses two completely separate things. GPL governs distribution of derivative works. An LLM trained on GPL code does not distribute that code. The model weights are not a cop
171.
by Rochus 4mo ago
I have many GPL projects (e.g. https://github.com/rochus-keller/Oberon , https://github.com/rochus-keller/Luon , https://github.com/rochus-keller/Micron ) and spend a significan
172.
by Rochus 4mo ago
Same here. The 2b, 3b and Zero 2 have the best cost-benefit ratio. I recently migrated the Oberon System 3 to those three boards and particulary the Zero is a great host for this bare-bone system: https://github.com/rochus-k
173.
by Rochus 4mo ago
When you consider SQLite, you might want to get around SQL for impedance and performance reasons. There is a robust and decently fast key/value store in the backend. I used it in some of my projects (e.g. https://github.com&
174.
by Rochus 4mo ago
Thanks, great that the work is useful.
175.
by Rochus 4mo ago
V8 was significantly faster than LuaJIT when I did measurements in 2021 (see https://github.com/rochus-keller/Oberon/blob/master/testcase... , didn't measure it since). Sure there are microbenchmarks
176.
by Rochus 4mo ago
Cool. It's still a great backend, even for a statically typed language, and the FFI and debugger features are very useful. Mono is yet another great backend which in addition supports multiple threads and has the higher performance, bu
177.
by Rochus 4mo ago
And a related paper by the same group: https://news.ycombinator.com/item?id=48262160
178.
by Rochus 4mo ago
Here is the repository with the benchmarks: https://github.com/rochus-keller/are-we-fast-yet/ And here is a repository with a compiler backend to generate LuaJIT bytecode and other useful tools: https:/&#x2F
179.
by Rochus 5mo ago
This is incredible. Wouldn't have thought that it works so well in the Firefox on my old Lenovo T480.
180.
by Rochus 5mo ago
Interesting results. Particularly they barely found a speed-up of newer compared to older LuaJIT versions (rather the contrary). Maybe they should have used the Are-we-fast-yet suite instead of the (random looking) set of microbenchmarks. I

More ›