On 6/28/11 12:26 AM, Allison Henderson wrote:

On 06/27/2011 10:09 PM, Dave Chinner wrote:

quoted

On Mon, Jun 27, 2011 at 09:27:26PM -0700, Allison Henderson wrote:

quoted

New filtered golden output for test 252

Signed-off-by: Allison Henderson<redacted>
---
:100644 100644 930c924... fcfd121... M    252.out
  252.out |  272 +++++++++++++++++++++++++++++++++++++++------------------------
  1 files changed, 170 insertions(+), 102 deletions(-)

diff --git a/252.out b/252.out
index 930c924..fcfd121 100644
--- a/252.out
+++ b/252.out

@@ -1,239 +1,307 @@
  QA output created by 252
      1. into a hole
+daa100df6e6711906b61c9ab5aa16032
      2. into allocated space
-0: [0..7]: data
+0: [0..7]: extent
  1: [8..23]: hole
-2: [24..39]: data
+2: [24..39]: extent
+cc58a7417c2d7763adc45b6fcd3fa024

I don't really like the way this weakens the test for XFS. With this
change, the test no longer is checking that unwritten extent
behaviour is correct.

Rather than weakening the test, perhaps it would be better to
execute 252 for XFS only (with the md5sums added), and then
duplicate it to a new test for all filesystems to run with the
weaker result checking that using the new filter function gives us.
With the amount of common code the two tests share, it should be
trivial to do this....

Alrighty, that sounds pretty straight forward, as long as every one
is in agreement. I think that would help retain the tests
effectiveness. Eric, Josef, what are your thoughts?

Yeah, I agree, I share Dave's concerns and that sounds like a good
way to go.

Thanks,
-Eric

Allison Henderson

quoted

Cheers,

Dave.