[master 2/4] Manage backing store more independently in loop backed test cases.

mulkieran installerbot-noreply at redhat.com
Tue May 26 12:58:57 UTC 2015


This is a standard worklist algorithm, it pulls stuff off the front of the list and pushes remaining work items onto the back. It's the 0 argument to pop() that makes all the difference. So, it really does give a minimum of 3 chances to every item in the list. I'll add an explanatory comment.

In practice, the fact that removeLoopDev() and os.unlink(store) are in the same try block does not cause a bug, because losetup returns success so long as loop device is detached, even if it was detached
already. But it is confusing, and the removeLoopDev() method may change, so I'll separate the two method calls.



-- 
To view this comment on github, visit https://github.com/mulkieran/blivet/commit/4f1671e812b8d0b8113520d55b89807fbf8686c7#commitcomment-11363360


More information about the anaconda-patches mailing list