On 10/16/2014 09:57 AM, Bernhard Voelker wrote: > On 10/16/2014 10:34 AM, Pádraig Brady wrote: >> On 10/16/2014 09:24 AM, Bernhard Voelker wrote: >>> The MHz number in /proc/cpuinfo may change, thus leading to >>> a false positive failure when comparing the expected against >>> the actual output file. >>> >>> * tests/misc/head-c.sh: Replace the actual MHz number in the files >>> to be compared by a placeholder. > >> Oh right good one. >> It's probably better to use an invariant file rather than >> messing with sed and GHz and possible other variant parts. >> >> How about /proc/version > > Well, /proc/cpuinfo is a multi-line file while /proc/version > is not. Maybe it's better to use another multi-line one which > does not change too often - like /proc/filesystems ?
That would add a little more inconsistency. Paul's patch used /proc/version elsewhere (which I just noticed now). Multi-line is irrelevant to this test, so I'd just use /proc/version here also. BTW, I just noticed a separate issue while looking at /proc usage in the tests, which the following should fix up. thanks, Pádraig. commit 437f24764963ca2935e1bce368667f8270275a4b Author: Pádraig Brady <[email protected]> Date: Thu Oct 16 10:33:46 2014 +0100 tests: fix test hang with unstable inodes in /proc * cp/proc-zero-len.sh: Search the 'err' file for the error to ignore, not stdin. diff --git a/tests/cp/proc-zero-len.sh b/tests/cp/proc-zero-len.sh index 3fcd5aa..8dd8460 100755 --- a/tests/cp/proc-zero-len.sh +++ b/tests/cp/proc-zero-len.sh @@ -31,7 +31,7 @@ cat $f > out || fail=1 # number of the file changed during the cp run. cp $f exp 2>err \ || { fail=1; - grep 'replaced while being copied' \ + grep 'replaced while being copied' err \ && skip_ "File $f is being replaced while being copied"; } # Don't simply compare contents; they might differ,
