Firm Bugtracker - libFirm
Viewing Issue Advanced Details
150 optimisations major always 2014-12-30 09:13 2016-01-03 03:28
Matze  
UniQ  
low  
closed development  
fixed  
none    
none  
0000150: missoptimized floatingpoint Add
ieee754 features negative zeros, that means (note that I do not use == in the following sentences as == ignored the difference between zero and minus zero):

x+0.0 is not the same as x in floatingpoint, because -0 + 0.0 results in 0.0.

x+(-0.0) however is always the same as x and can be optimized to x.

There is a testcase in firm-testsuite/opt/minus_zero.c now.
Issue History
2014-12-30 09:13 Matze New Issue
2014-12-30 09:13 Matze Status new => confirmed
2015-05-08 13:46 UniQ Status confirmed => assigned
2015-05-08 13:46 UniQ Assigned To => UniQ
2015-05-08 13:47 UniQ Note Added: 0000226
2015-05-08 13:47 UniQ Status assigned => resolved
2015-05-08 13:47 UniQ Resolution open => fixed
2016-01-03 03:28 Matze Status resolved => closed

Notes
(0000226)
UniQ   
2015-05-08 13:47   
Fixed in libfirm 59530b3544e291d3e7ac38cd8d3d8b8b75a3ea39.