On Jul 2, 2013, at 10:28 AM, Koen De Wit <koen.de.wit@xxxxxxxxxx> wrote:
> Dave,
>
> Thanks for the review. I will clean up the commit message and do a full mail-to-myself-and-test-patch round trip to avoid errors like the wrong test numbers in the golden output. I'm sorry for this.
>
> About cutting out file names from the output. I did this in the first version of the patch:
>
> md5sum $TESTDIR1/$F | $AWK_PROG 'END {print $1}'
>
> but Eric Sandeen suggested to include them in order to provide more context in the output. (See http://oss.sgi.com/archives/xfs/2013-03/msg00231.html and http://oss.sgi.com/archives/xfs/2013-03/msg00220.html) That sounds like a good idea to me, it makes debugging failures easier. Whose opinion should I follow?
>
Heh sorry. IMHO maybe a middle ground; not bare md5sum but show only the base name? In the end up to you; it seems Dave and I have different opinions on this. :)
Eric
> Koen.
>
>
> On 07/02/2013 12:15 PM, Dave Chinner wrote:
>
>> On Tue, Jul 02, 2013 at 11:27:51AM +0200, Koen De Wit wrote:
>>> This testscript creates reflinks to files on different subvolumes, overwrites the original files and reflinks, and moves reflinked files between subvolumes.
>>>
>>> Originally submitted as testcase 302, changes are made based on comments from Eric: http://oss.sgi.com/archives/xfs/2013-03/msg00231.html
>>> Two new common/rc functions used in this script (_require_cp_reflink and _verify_reflink) have been submitted recently: http://oss.sgi.com/archives/xfs/2013-05/msg00745.html
>>> Thanks to Eric Sandeen and Dave Chinner for the reviews.
>>>
>>> Version 3: fixing wrapped patch.
>>
>> Needs a proper commit message - wrap it at 72 columns, change log
>> should be placed below the --- devider, not be part of hte commit
>> message.
>>
>> Test numbers do not need to be unique across all test directories,
>> just unique within the tests/btrfs directory.
>>
>>> +
>>> +_checksum_files() {
>>> + for F in file1 file2 file3
>>> + do
>>> + for D in $TESTDIR1 $SCRATCH_MNT $SUBVOL2
>>> + do
>>> + md5sum $D/$F | _filter_test_dir | _filter_scratch
>>
>> Just cut the file name out. No need for filtering at that point...
>>
>>> + done
>>> + done
>>> +}
>>> +
>>> +TESTDIR1=$TEST_DIR/test-$seq-1
>>> +TESTDIR2=$TEST_DIR/test-$seq-2
>>> +SUBVOL1=$TEST_DIR/subvol-$seq-1
>>> +SUBVOL2=$TEST_DIR/subvol-$seq-2
>>
>> ....-316-.....
>>
>>> +Verify the file contents:
>>> +e09c80c42fda55f9d992e59ca6b3307d TEST_DIR/test-302-1/file1
>>
>> which means the golden output is broken, despite the filtering
>>
>> You should test your patches before posting ;)
>>
>> FWIW, this is why you should simply cut the filename completely out
>> like we do elsewhere....
>>
>> Cheers,
>>
>> Dave.
>
>
_______________________________________________
xfs mailing list
xfs@xxxxxxxxxxx
http://oss.sgi.com/mailman/listinfo/xfs
