Inlining too conservative
|
|
Bugzilla Link |
426 |
Created on |
May 10, 2008 20:41 |
Resolution |
INVALID |
Resolved on |
May 13, 2008 16:43 |
Version |
1.00beta |
OS |
Linux |
Architecture |
PC |
Extended Description
If I compile sac/testsuite/optimizations/constantfolding/CFwith.sac
with -ecc -b11 -doinl, it fails to inline the trivial "MAIN::add"
function. This causes a CF optimization (0+X ---> X) to get missed.
If I EXPLICITLY mark the function: "inline MAIN::add", it does
get inlined, and things work properly.
So, either inlining is not working properly, or the inlining capability
is inadequate documented for Canadian readers.