I guess what I’m asking is that is there a limit that packages will hit when needing to optimize for better performance while being maintainable like what happened to LoopModels
. This specific package was just the context in which I first saw this happen.
That’s his opinion and he’s welcome to it. Most people here disagree with him (including myself) and that’s okay.
Some people really like C++. Some people don’t. You’ll find a lot more people here who would prefer to not write library code in C++ than you would the other way around. If you went to a C++ forum, you’d likely find that most people there prefer to write libraries in C++, rather than julia. Surprise surprise.
There’s tradeoffs here. Neither language is uniformly better at everything.
Sidenote, can we please not make this conversation about someone’s opinion who clearly doesn’t want to talk about this here? I regret making that comment in the first place, as it wasn’t really my place to say it anyways.
Maybe a moderator can just delete my old comment and edit it out of the quotes here in this thread (cc @mbauman @gdalle?)
ah I’m sorry, I didn’t realize it was inappropriate. I’d be happy to delete this thread (if discourse allows) and maybe make a new one without any quotes? I just quoted for context and because (I assume) everybody mentioned knows a lot more than me.
It’s definitely leaning towards the kind of internet drama that isn’t productive — through no fault of your own here you’re quoting someone quoting from a private DM speculating about someone’s internal motivation in a short enough quip that it’s basically a hot take.
That’s probably not a great start for a thread, no.
I do think the general question itself is fine — especially if it is looking at actions instead. E.g.,
I see that LoopModels says that it’s “intended to be the successor to LoopVectorization.jl”
It’s written in C++. Is there a cliff upon which packages will hit like what happened to LoopModels?
While I do have the power, I’m very loathe to edit others’ posts. It’d be great if you would do so, however, especially as Mason requested it. I can then move the subsequent commentary about the direct quotes into a separate — potentially unlisted — topic.