Re: bash's test and GNU test give different results [if file not there]

[Dan Jacobson]

Chet says I should file a bug on gnu.utils.bug.

[If you ask me, this will make GNU test "defective" like bash test,
but that is just me... anyway, I leave the solution to you pros.
I still hope to see lots of warnings in the manpages.]

>>>>> "Chet" == Chet Ramey <address@hidden> mailed me:

>> >>>>> "Dan" == Dan Jacobson <address@hidden> writes:
>> 
Dan> $ test non_existant_file -ot . && echo preposterous
Dan> preposterous
Dan> $ /usr/bin/test non_existant_file -ot . && echo preposterous
Dan> $ test . -nt non_existant_file  && echo preposterous
Dan> preposterous
Dan> $ /usr/bin/test . -nt non_existant_file  && echo preposterous
Dan> $ echo $BASH_VERSION
Dan> 2.05.0(1)-release
>> 
>> >>>>> "Chet" == Chet Ramey <address@hidden> writes:
>> 
>> >> Erm, so did I find a big bug or not?  Thanks.
>> 
Chet> I don't think so.  Non-existent files have their mtimes set to 0.
>> 
>> OK, then what are we going to do about the different results test -ot, -nt
>> give for bash's test vs. the GNU test?
>> 
>> Shall I repost to gnu.utils.bug instead?

Chet> Sure, file a bug report.  The `-[x]t' tests originated with ksh, so
Chet> bash is compatible with ksh93.  There's no reason to be deliberately
Chet> incompatible here.
-- 
http://www.geocities.com/jidanni/ Taiwan(04)25854780