And thus demonstrated is the value of lots of different implementations of a spec. Already one hole found in the spec in just this article, and I'm sure there will be/were more.
I love that this project keeps showing how possible it is for a small group to make something amazing. This would be very hard to do in a company with stakeholders.
For issue #3, it might also be a good idea to have a maxdepth mechanism in gradients that point to other gradients; this would be a defense in depth control vs some error or limitation in your “have I seen this reference before” logic. I’m not familiar with SVG gradients; maybe there is a reason to have reference chains of these 1000 links long, but I’d bet that if you ever encounter this in the wild then it’s an attack or a fuzzer.
This comment is being left from Ladybird. Hacker News works in Ladybird now. I use Ladybird for the few minutes a day that I surf sites like Hacker News and OSnews.<p>It is slow. It is fragile. But it works. That alone is amazing given how young the project is and how they have written literally everything from scratch.<p>I am really looking forward to Ladybird maturing.
Interesting thanks.
What bothers me though is that almost all developers do exactly what you see in issue #1: We found it! fix committed done!
Nope, you should understand exactly what went wrong: assuming parents must exist... Now search the entire codebase for the same kind of mistakes. Use your creative brain to figure out where else same thing can happen. It will never be in just done place.
All modern software is unreliable bug ridden nightmare, mostly because of capitalism constraints yes... but it is possible to do better