Adding a useless function is a non-breaking change. Not comparable to breaking an existing one.
Agreed. But for most users reproducibility on their own system sufficed. If sufficiently important on another you could LD_PRELOAD a replacement.
-
-
Bwahahaha Okay, so you admitted that predictable rand() is just a local convenience, and you're now advocating a dirty hack to make it portable. If you need reproducibility in a production setup, you shouldn't be using rand() anyway.
-
No, it makes it *possible* to reproduce the results you got once, even if a lot of work is involved. With csPRNG where it doesn't belong, it's impossible to ever reproduce.
-
Right. That's why we added *_deterministic versions. Better to have to explicitly ask for *deterministic versions than to have !csPRNG by default.
-
We have standards and specifications for a reason...
-
but sometimes, the standards process gets hacked, or too slow, or bad. Older story \oe isn't in ISO Latin 1, because the french representative was from Bull, whose printers didn't do \oe at the time There are definitive problems with the C standard. Take restrict, for instance.
-
Individual standards are very flawed yes. But the principle of (flawed) consensus, common base expectations, avoiding vigilante changes, etc. is still worth a lot.
-
there's a chicken & egg issue at work. Sometimes you do change stuff because then you hope to force the standard.
End of conversation
New conversation -
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.