What do people think of me adding something called hack() that behaves
like warn() or warn_once(), but is only compiled in when a HACK_ON
flag is set (by default debug/opt) and is turned on by a command line
option?
The idea here is that right now, we have no consistent way of
indicating that something needs work and is a hack. This way, we can
run an experiment with the hack flag on and get some information about
what hacks were encountered during the simulation experiment.
Would people use it? Would people actively try to convert "hacks" to use it?
sample usage would be:
int
foo()
{
hack("this function should really check the return value of write and");
hack("signal an error if necessary, but we ignore that for now");
write(fd, "asdf");
}
Nate
_______________________________________________
m5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/m5-dev
