On May 28, 2008, at 10:38 AM, nathan binkert wrote:
Ok, I'll admit, I still don't remember the difference between panic
and fatal. I pretty much always use panic, but I think I should be
using fatal sometimes. What was the difference supposed to be? Can
we rename one of them?
panic() -- An assert that isn't compiled out
fatal() -- some configuration parameter caused a problem
I think one of them was supposed to be a user error and the other was
supposed to be a simulator error (similar to an assert?). I started
thinking of this because the "user errors" like the "stat check
failed" error should probably all have some sort of descriptive error
message somewhere.
Can't we make the error messages themselves descriptive, rather than
requiring a two step process?
If we (I'm willing to put effort into this) try to
get these separated well, we can start a wiki page for the "user
errors" and actually have the simulator spit out a URL that points to
the error page. I'm willing to also create a python equivalent for
the user error. I think we should do more try/except blocks and throw
a user error so the user doesn't just see a python stack trace (unless
they ask to by setting a command line option).
Most of these errors panic in the c++ and call _exit(). I don't think
people are
seeing python error messages that often.
Ali
What do people think? If I put the mechanism in place, can we divide
up the code tree and each tackle some section? I'll volunteer to do
src/sim, src/python, and src/base/stats. I think those would contain
the majority of user errors.
Nate
_______________________________________________
m5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/m5-dev
_______________________________________________
m5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/m5-dev
