On Feb 11, 2008, at 9:02 PM, Ruediger Pluem wrote:
This non fatal here and may happen due to races with htcachelean. So
I wouldn't
check the return value here and let it fail later in
safe_file_rename if it
really doesn't work.
Ok - I'll remove the checks for now and will give it a play -- I
introduced them as I found
them helpful exactly there when playing with intentional disk-full and
unmounting of
the cache disk* -- and finding the old behaviour very mistyfiing.
Why shouldn't we remove the tempfile in this case?
Mistake - fixed !
Dw.
*: temp volume in memory/swapspace.
