Yes, any time behavior is changed, either to fix a bug or introduce a new 
feature, there is the possibility that someone somewhere might have been 
relying on the previous behavior. But you can't introduce backwards compatible 
settings for each and every change ever made, that would just be unwieldy and 
poor design.

In this case, the change is restoring proper behavior for a documented 
interface that was temporarily doing the wrong thing. Particularly given as of 
yet there haven't been any concrete examples as to how this might be desirable 
behavior, this doesn't seem like a scenario where such a kludge would be 
appropriate. I take it you work for RackTop? Why is it you think your customers 
would want to continue with the broken behavior rather than the correct 
behavior for this aclinherit setting?

Further, as I suggested, if this behavior is desirable for someone, it would be 
better implemented as a separate aclinherit option rather than providing the 
confusing ability to have an existing documented mode behave differently. The 
current broken behavior wouldn't be the default, so an admin reading the 
release notes would have to do something to maintain it, whether it be setting 
a kernel tunable or changing an aclinherit setting. Plus a new setting would 
allow it to be set on a per filesystem basis rather than globally; what if you 
have some customers who want the fixed behavior and others who want the 
"different in general but broken for this setting" behavior?

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/openzfs/openzfs/pull/557#issuecomment-367145972
------------------------------------------
openzfs-developer
Archives: 
https://openzfs.topicbox.com/groups/developer/discussions/Taa24d3ad3ace2410-M5fdc823746043a62f3518ab9
Powered by Topicbox: https://topicbox.com

Reply via email to