wrapping signed integers (vs undefined signed overflow) slow down SPEC CPU 2006 by 1.4% when compiled by Clang 3.8 cc @samth
-
-
Replying to @johnregehr
@johnregehr makes the undefinedness seem pretty inexcusable, doesn't it?4 replies 2 retweets 2 likes -
Replying to @johnregehr
@samth but more seriously, the UB makes checkers like UBSan possible -- you can't do that in Java2 replies 0 retweets 3 likes -
Replying to @johnregehr
@johnregehr Why not? It's more restrictive so has to be opt-in, but so is UBsan.1 reply 0 retweets 0 likes -
Replying to @johnregehr
@johnregehr Surely UBsan is a dialect as well, in that you wouldn't be able to automatically switch it on in production.1 reply 0 retweets 0 likes -
Replying to @johnregehr
@johnregehr My point is that "not broken according to the C standard" is a dialect that most people don't program in.2 replies 0 retweets 4 likes
Replying to @samth @johnregehr
The people who refuse to program in it need to stop claiming they are programming/can program C. They're not.
9:22 AM - 14 Nov 2016
0 replies
0 retweets
0 likes
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.