Dan Gohman d479449624 Disable the (name+num)+num = name+newnum optimization. ... This optimization reassociates the addition and can introduce situations where intermediate computation wraps around the address space, creating inefficiency in JS engines which are optimized for no-wraparound cases. In its place, the fastcomp backend is being enhanced to fold getelementptrs with constant bases, which is the main case where the above optimization happens in practice.		2015-05-06 10:28:20 -07:00
..
CMakeLists.txt	Add high level profiling for optimizer stages	2015-04-22 14:17:25 +01:00
istring.h	…
optimizer.cpp	Disable the (name+num)+num = name+newnum optimization.	2015-05-06 10:28:20 -07:00
parser.cpp	…
parser.h	…
simple_ast.cpp	Remove useless constructor parameter	2015-04-18 02:51:58 +01:00
simple_ast.h	…
snprintf.h	…