The code uses GetRandomBytes in several contexts where it's either unwieldy to return an error, or an error would mean a fatal exception anyhow. Bitcoin can't cope gracefully with a failing RNG. So just fail with an assertion error. @gmaxwell checked OpenSSL a while ago and discovered that it never actually fails, but it can't hurt to be a bit paranoid here.
(somehow this gives me a deja-vu, I remember we did this before? can't find it in git history though)
We're using GetRandomBytes in several contexts where it's either
unwieldy to return an error, or an error would mean a fatal exception
anyhow.
@gmaxwell checked OpenSSL a while ago and discovered that it never
actually fails, but it can't hurt to be a bit paranoid here.
Increases our reliance on being compiled with assert() for correct behavior in some abstract sense. I'm not sure if I care given that it will not fail.
utACK - we're already relying on assert, and test for it. If we ever want to get rid of it, it's obvious where in code changes are needed.
ut ACK
utACK
Tested ACK only commithash 65e3a1e76202b9a695fc6319dbd527ac563e0895: http://bitcoin.ninja/TheBlueMatt-5239.txt
Right - we need to replace asserts with a macro that calls to a custom function that cannot be compiled out at some point.