??? 07/05/08 06:53 Read: times |
#156406 - Usually not Responding to: ???'s previous message |
This has nothing specifically to do with the 8051 - so should really be in Chat...
Rm Nair said:
If the unoptimal code runs correctly, but the optimized code does not, is not the defect in the compiler? Usually not. Usually this means that the code is faulty, and only works by luck in the unoptimised version; eg, it may be based on some false assumption that just happens not to fail when unoptimised Should not the compiler vendor fix the defect, and in the interim provide a work around? In the rare cases that it is a compiler fault, then yes - of course they should! NB: in the case of an open-source compiler, who do you consider to be the "vencor"...? |
Topic | Author | Date |
Why Is Debugging Optimized Code Such an Issue? | 01/01/70 00:00 | |
Usually not | 01/01/70 00:00 | |
False Assumptions | 01/01/70 00:00 | |
"language specification" ?? | 01/01/70 00:00 | |
Do not assume it is that easy | 01/01/70 00:00 | |
Common Stuff Leading To Errors | 01/01/70 00:00 | |
"common stuff" falling apart when optimized | 01/01/70 00:00 | |
How broken things work | 01/01/70 00:00 | |
It is not like that | 01/01/70 00:00 | |
A bad workman blames his tools | 01/01/70 00:00 | |
A common side effect of optimisation | 01/01/70 00:00 | |
Newbies | 01/01/70 00:00 | |
another good reason | 01/01/70 00:00 | |
a rule of thumb | 01/01/70 00:00 | |
excellent advice, however - and | 01/01/70 00:00 | |
Defeatist? | 01/01/70 00:00 | |
Or turn it down as needed | 01/01/70 00:00 | |
Using Pragmas During Debugging | 01/01/70 00:00 | |
Free lunch? | 01/01/70 00:00 | |
a dream | 01/01/70 00:00 | |
If it were easy everone would be doing it. | 01/01/70 00:00 | |
you can't | 01/01/70 00:00 |