bug-coreutils
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Debian Bash bug #173273 (test -w with no arguments isn't an error)


From: Dan Jacobson
Subject: Re: Debian Bash bug #173273 (test -w with no arguments isn't an error)
Date: 19 Dec 2002 05:51:09 +0800
User-agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.2

>>>>> "Chet" == Chet Ramey <address@hidden> writes:

Chet> I don't understand it.  Do people deliberately ignore the relevant 
sections
Chet> of the documentation so they can claim it doesn't cover something?

Chet> This is from the man page, in the section describing `test':

OK, but you had better add something about it to the "help test" output.

Chet>               test and [ evaluate conditional expressions using a
Chet>               set of rules based on the number of arguments.

Chet>               0 arguments
Chet>                      The expression is false.
Chet>               1 argument
Chet>                      The expression is true if and  only  if  the
Chet>                      argument is not null.

OK, but you must admit that without something about this being
mentioned in the 'help' output, the user assumes it is like the other
99.9% of the commands.   I mean what if the rm command was all OK
about any single switch with no arg being not an error...

anyway, take a 2nd look at the help output. it sure needs a note about
any single unnil arg returning true.

And the GNU test(1) man and Info pages  don't mention any of that.
That is surely their bug.

Chet>               2 arguments
Chet>                      If  the  first argument is !, the expression
Chet>                      is true if and only if the  second  argument
Chet>                      is  null.   If  the first argument is one of
Chet>                      the unary conditional operators listed above
Chet>                      under  CONDITIONAL  EXPRESSIONS, the expres-
Chet>                      sion is true if the unary test is true.   If
Chet>                      the first argument is not a valid unary con-
Chet>                      ditional operator, the expression is  false.
 
Chet>                   [...and so on...]

Chet> How does this fail to describe the case where test is given a single
Chet> argument of `-w'?

well, only one [bash man] of the 4 references [bash man, bash help,
man test, info test] one would check mentions this. 

Chet> This is explicitly covered in the FAQ, question E1.

but it is more important the meanings of -w -f -d etc. so it should be
mentioned first everywhere.
-- 
http://jidanni.org/ Taiwan(04)25854780



reply via email to

[Prev in Thread] Current Thread [Next in Thread]