Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
p0nce
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
37 ms
·
271.
▲
by
p0nce
5y ago
> those who [used the GC] required only about a third as much time (4 hours vs. 12 hours). This makes me happy to use D instead of Rust. D has a @safe subset.
272.
▲
by
p0nce
5y ago
Quoting the article: > those who [used the GC] required only about a third as much time (4 hours vs. 12 hours). Surely you can devote a chunk of your newfounded time to find use-after-free bugs.
273.
▲
by
p0nce
5y ago
Perhaps some kind of centralization could help with that.
274.
▲
by
p0nce
5y ago
When porting our audio plug-ins to Big Sur we had to avoid some APIs that were running fine for years, because of new memory leaks.
275.
▲
by
p0nce
5y ago
Have you used both?
276.
▲
by
p0nce
5y ago
We welcome people that want to make a plugin business foremost (or already are in the market). https://dplug.org/
277.
▲
by
p0nce
5y ago
I much more happy using D that I have ever been using C++. It is a perfect replacement, with none of the issues mentionned in GP comment.
278.
▲
by
p0nce
5y ago
Leaving is the professional thing to do.
279.
▲
by
p0nce
5y ago
+1 Every time I tried --fast-math made things a bit slower. It's not that valuable with LLVM.
280.
▲
by
p0nce
5y ago
I learned with QBASIC because you could put text and pixels on screen immediately. It is beyond useless to start with syntax, fibonacci and prime numbers.
281.
▲
by
p0nce
5y ago
I have no DB, static website, deploying in production on a VPS with git (manually) and rsync + Cloudflare. I just don't know what kubernetes is, nor do I encounter any of the problems you mention. > Just properly setting up and main
282.
▲
by
p0nce
5y ago
Climate is getting warmer so we need humans that can stand higher temperatures.
283.
▲
by
p0nce
5y ago
Is it Stockhold syndrome? After years of GIMP I really disliked the Affinity interface, also it is unusably slow on a very recent computer here. I kinda like GIMP, the good parts.
284.
▲
by
p0nce
5y ago
For me the problem is that "functional programming" embodies a lot of questionnable ideas: - the idea that readability is not the first thing to pursue - the idea that macros are great - the idea that leaving performance off the t
285.
▲
by
p0nce
5y ago
Apple contributed hugely to kill Flash and Java on the Desktop (let alone things like Silverlight). ALso non-native code is still 2x slower all things considered.
286.
▲
by
p0nce
5y ago
Qt, GTK, SDL, Flutter...
287.
▲
by
p0nce
5y ago
It is more CPU intensive than a baseline JPEG.
288.
▲
by
p0nce
5y ago
Decoding a progressive JPEG when you have the whole stream is more CPU intensive and the decoder itself is more complex.
289.
▲
by
p0nce
5y ago
What is the downside?
290.
▲
by
p0nce
5y ago
Yes. But this role doesn't always exist. For whoever is in power, creating suitable counter-powers may not feel like a pressing issue.
291.
▲
by
p0nce
5y ago
There is a dark organizational pattern that I find wasteful: when sales have complete power over engineering, sales would routinely ask for features/projects at a breakneck pace for _potential_ customers instead of having a product-bas
292.
▲
by
p0nce
5y ago
Sounds like a headline optimized for HN success!
293.
▲
by
p0nce
5y ago
> the fun, gamery loading messages with something more "formal". It may be worse for you, but it feels way better to me. It has been very annoying to introduce Discord in work/casual settings and see it labelled as "a
294.
▲
by
p0nce
5y ago
D has the @nogc attribute. Using it to make memory allocation forbidden in parts of the codebase.
295.
▲
by
p0nce
5y ago
Website: HTML/CSS/JavaScript, hosted with Nginx on one VPS. D language for products/tools/website. SaaS: - CloudFlare - Google My Business for emails
296.
▲
by
p0nce
5y ago
Hacker News skews towards richer, more technical, and higher-educated and then makes systematic bias against pedestrian, working-class, and uncomplicated solutions. If it's too popular on HN I'd say it's a red flag.
297.
▲
by
p0nce
5y ago
Static types are small-time unittests so no.
298.
▲
by
p0nce
5y ago
Instead of 2: Multiply by π for new kind of projects. Multiply by φ for known projects.
299.
▲
by
p0nce
5y ago
> they gain domain knowledge that is specific to that company, and that stuff is incredibly valuable. The wild thing is that it’s only valuable to that one company! From the engineer point of view, this is a bad deal as the knowledge the
300.
▲
by
p0nce
5y ago
I don't buy the thesis that everydone was doing software development wrong until a few consultants gathered at a ski resort.
More ›