| |||||||||||||
Workqueues and internal API conventionsWorkqueues and internal API conventionsPosted Aug 31, 2006 21:57 UTC (Thu) by giraffedata (subscriber, #1954)Parent article: Workqueues and internal API conventions
This seems to fly in the face of Linux kernel convention. The policy has always been, "keep in-tree code clean; screw out-of-tree code." I can't count the number of times that policy has been implemented. Here, we have in-tree code dirtied by having two functions for the same thing, for the purpose of backward compatibility with some out-of-tree code. I wonder why there wasn't objection.
(Log in to post comments)
Workqueues and internal API conventions Posted Sep 6, 2006 19:19 UTC (Wed) by roelofs (guest, #2599) [Link] I wonder why there wasn't objection.There's a huge difference between compile-time failures and silent runtime failures. Linus has made it very clear that he will not accept other forms of "functionality regression" at the end-user level, and I suspect this is a logical expression of that same pragmatic philosophy. Greg
|
|||||||||||||