3 ms·
It's a very biased and old list of articles from an author who openly claimed that "V should die". For example, the performance of V is measured on a debug bui
by amedvednikov 5y ago
It's a very biased and old list of articles from an author who openly claimed that "V should die".
For example, the performance of V is measured on a debug build, without vlib cached, without vfmt disabled, with a 10x slower backend.
Here's V compiling itself in 0.3 seconds:
https://www.youtube.com/watch?v=pvP6wmcl_Sc https://www.youtube.com/watch?v=pvP6wmcl_Sc
Also complaining about not running on all Linux flavors on day 1. Right now V runs on literally everything.
And so on.
That's why I asked to list the specific claims the poster had in mind here in the comments.
- pwdisswordfish9 5y ago> an author who openly claimed that "V should die". Citation needed. Or is the source for that claim the same as that for V benchmarks?
- amedvednikov 5y agohttps://news.ycombinator.com/item?id=27442724 https://news.ycombinator.com/item?id=27442724 > I think it is something that should be ignored until it dies into obscurity