undefined | Better HN

0 pointsAnimats2y ago0 comments

> They've also written an article about reflection in Rust - a purely technical thing that is already pretty widely disliked conceptually.

Maybe that really is the problem. Too much reflection leads to clever, but hard to debug hacks. Look what happened with C++ templates and, further back, LISP macros. That way lies code that's unusually hard to read and maintain. Rust probably shouldn't go that way.

Do you want to have a keynote address on an idea of that type? It's more of a subject for a proposal talk.

0 comments

lolinder2y ago

The would-have-been speaker's blog post addresses this question specifically. He apparently pushed back many times on the idea of giving the keynote on his work and was assured that the Rust organization was aware of the topic and wanted him to go ahead [0]. Then they walked back on that.

That's what I think most people find terrible about Rust leadership's handling of the situation. It's not that this specific keynote was super important, it's that the team let the speaker spend a few weeks preparing a keynote and then pulled the rug on him.

[0] https://thephd.dev/i-am-no-longer-speaking-at-rustconf-2023

shash2y ago

C++ templates are not reflection; either compile time or runtime. There is a form of reflection, but that's RTTI, which is a totally different ballgame.

AnimatsOP2y ago

True, but both come under the heading of adding a level of indirection at compile time. It's a tough language decision on how far to go in that direction. Go, pre-template, was minimalist on this. C++ probably went too far. Where to stop is a good question.

j / k navigate · click thread line to collapse