On 30/08/15 22:04, Bernhard Voelker wrote:
> On 08/29/2015 03:14 AM, Pádraig Brady wrote:
>>> As we can't predict the number of stat() calls as before,
>>> what shall we do to fix the test?
>>
>> Maybe something like:
>>
>>   -strace -o log-help -e $stats ls --help >/dev/null || fail=1
>>   -n_lines_help=$(wc -l < log-help)
>>   +strace -o log-init -e $stats ls . >/dev/null || fail=1
>>   +n_lines_init=$(wc -l < log-init)
>>
> 
> Thinking about it more, the test should verify that ls(1) doesn't
> need more stat-like calls for a populated directory argument (with
> regular files, symlinks, etc. below) than for an empty directory
> argument, shouldn't it?
> 
> The new version of the test in the attached patch does exactly this.
> 
> Thanks & have a nice day,
> Berny
> 

+1

thanks,
Pádraig.


Reply via email to