Abdelrazak Younes wrote: > Alfredo Braunstein wrote: >> Abdelrazak Younes wrote: >> >> >>>> But it's the least helpful and the most annoying approach. >>>> We should change that. >>> Sure but for trunk. >> >> I agree with Andre'. We should abort ASAP with an emergency file, with >> e.g. memory corruption. So IMO asserts are fine in the final release. > > I am not sure but that's not Andre's point...
It seemed to me, maybe I was wrong. > I disagree because there are cases (such as this one) where we use > asserts instead of handling the errors gracefully. We should avoid > aborting as much as possible. I agree, but this means we should substitute asserts with graceful error handling when possible. It doesn't mean we should mass-disable asserts in the final release. IMO the risk of dataloss by *ignoring* an important assert is much worse than the eventual gain of ignoring unimportant ones. A/
