so replacing a division by 0 with an assertion? what's the difference? (SIGFPE vs SIGABRT?)

i did do a similar thing in redis recently, and my aim was for the tests to be able to easily distinguish between a crash by signal, and an assertion (also added use-exit-on-panic config to silence valgrind warnings).

my point is, if this is a completely dead code, why bother?
if it's not dead, we may wanna kill it (add a check earlier with a proper error message).
or we can also say we don't care much claiming that it's an invalid use of the tool (a cli tool, not redis server).

It can't happen and an assert is just a way to make that more clear in the code. If the function starts to be used in another place, maybe it can happen. @oranagra if you think it's useless, feel free to close the PR and the issue. :-)

If the warning is from an analysis tool we want to use, we'd want to silence the false warnings. Some tools out there are quite good, but I don't know which one this was. @yiyuaner which tool was it?

It can't happen and an assert is just a way to make that more clear in the code. If the function starts to be used in another place, maybe it can happen. @oranagra if you think it's useless, feel free to close the PR and the issue. :-)

If the warning is from an analysis tool we want to use, we'd want to silence the false warnings. Some tools out there are quite good, but I don't know which one this was. @yiyuaner which tool was it?

It's from an ongoing research project. I don't know the coding practice of redis, but maybe this warning reveals at least dead code (if not potential bugs). Maybe an explicit assert(false) can help to clarify the intention of the programmer. Also, in this case it can also help to suppress the warning from the static analyzer.

i only not notice the title of the PR (i think i didn't noticed before that it mentions the reason is to silence a warning)
thank you.