I’m a believer in “if your infrastructure isn’t working, plan A should be to improve your infrastructure; elaborate workarounds should only be a plan B”.
-
Show this thread
-
This is how we developed rustc. We were always ready to develop patches for LLVM when it wasn’t working for us instead of working around it.
2 replies 2 retweets 12 likesShow this thread -
It’s also how I think about the Web graphics stack. I’m not a fan of the status quo of using a “compositor” to work around a subset of 2D graphics performance problems. The WebRender approach is to fix the entire stack instead.
1 reply 3 retweets 20 likesShow this thread -
And I think of GC the same way. I’m skeptical of attempts to carve out GC-free subsets of languages, heavy reliance on escape analysis, etc. Improve your GC first before working around it.
2 replies 1 retweet 12 likesShow this thread
I think it’s different when you design the entire system around no-GC from the get-go. (Besides, uniqueness was first intended as a concurrency safety measure first; the no-GC stuff came later.)
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.