baselayout-1 complains if an init script uses `exit' in a subshell, e.g.: ( do-some-env-mangling; exit 0 ) I don't see a reason why an init script isn't allowed to do that. I'd suggest removing that pointless, poor man's check.
that is an unfortunate side effect, and one i'm not sure can be fixed. the check was added because it was catching people wrongly using exit, and i imagine has continued to catch wrong code, so no, it is not "pointless".
It is pointless as it not only results in false positives but throws these at users instead of devs. So, we're getting into vicious circle where users keep reporting the same false positive just because some check may find a single broken init.d script somewhere.
users only see it when devs dont fully test their code. there is no need for you to use `exit 0`. a simple `:` at the end will work just as well. ( do-some-env-mangling; : )
And devs have to do a new release of a program just because some broken code says so. Not to mention that your example works only for a simple case like described here; with a more complex code you have to refactor everything just to get rid of an 'exit' at a random point.
... except the dev in question shouldnt have made said release in the first place as they should have seen the warning on their own systems if the code is so complicated to make missing exit a problem, then it sounds like the code is fundamentally broken. init.d scripts shouldnt be jumping through hoops.
(In reply to comment #5) > ... except the dev in question shouldnt have made said release in the first > place as they should have seen the warning on their own systems Assuming all the devs are obligated to use both versions of baselayouts at the same time. > if the code is so complicated to make missing exit a problem, then it sounds > like the code is fundamentally broken. init.d scripts shouldnt be jumping > through hoops. The point is that the `exit' call doesn't actually occur in the init.d script. It occurs in a sourced, external script. And I can make it not use it -- but why should I? Because some pointless, broken check tells me to?
hmm, lets see ... your decision to source an arbitrary external script of unknown contents in an init.d instead of executing things yourself is your problem. the check in question has existed for almost 6 years at this point. you're the first to complain about it. which makes me think that it's really not an issue and is actually useful. you know, because it has found real bugs. and as for not testing against multiple versions, it sounds like you're not fully testing against a stable system before stabilizing a package. yeah, like that's going to win you an argument. i imagine openrc through its large rewrites lost this check, so i can see about getting that restored if it's still feasible.
