[[ and [

[[ and [

[sergio]

Hello.

It's dumb question, but I can't find this in man.

What it the difference between [[ ]] and [ ]?
if [[ ]]; then ... fi
and
if [ ]; than ... fi

OK, the first is an evaluating of the conditional expression, and what
is the second?

-- 
sergio.

Re: [[ and [

[Bart Schaefer]

On Mar 4,  6:36am, sergio wrote:
}
} if [[ ]]; then ... fi
} and
} if [ ]; than ... fi
} 
} OK, the first is an evaluating of the conditional expression, and what
} is the second?

An alias for the "test" builtin -- so everything that follows "[" is
parsed/expanded like an ordinary command argument.  It's in the zsh
manual under the "test" command description.

Re: [[ and [

[Kurtis Rader]

[-- Attachment #1: Type: text/plain, Size: 1101 bytes --]

To expand on Bart's reply the "if [ ]" form goes back to the original
Bourne shell (probably before you were born but not before I was born :-)
which only supported invoking external programs and evaluating their exit
status. The left square-bracket was actually a filesystem alias for
/bin/test (or /usr/bin/test). Although today it is more likely that both
the "[" and "test" commands are built-ins to the shell that exhibit the
historic semantics. Whereas "[[" is a token that triggers the shell to
parse the contents up to the matching "]]" using different rules. I believe
the "[[ ]]" notation was introduced by the Korn shell after David Korn who
wrote the first version.

On Sat, Mar 3, 2012 at 6:36 PM, sergio <mailbox@sergio.spb.ru> wrote:

> Hello.
>
> It's dumb question, but I can't find this in man.
>
> What it the difference between [[ ]] and [ ]?
> if [[ ]]; then ... fi
> and
> if [ ]; than ... fi
>
> OK, the first is an evaluating of the conditional expression, and what
> is the second?
>
> --
> sergio.
>



-- 
Kurtis Rader
Caretake of the exceptional canines Einstein and Chino

Re: [[ and [

[sergio]

On 03/04/2012 08:24 AM, Bart Schaefer wrote:

May be all [ ] should be changed to [[ ]] in zsh distribution if it's
more native way?

% grep -R 'if \[[^\[]' /usr/share/zsh/* | wc -l
35


-- 
sergio.

Re: [[ and [

[Bart Schaefer]

On Mar 6,  4:32am, sergio wrote:
}
} May be all [ ] should be changed to [[ ]] in zsh distribution if it's
} more native way?

There may be some that could be changed, but they do have different
parsing rules so they aren't necessarily interchangable.

Re: [[ and [

[sergio]

On 03/06/2012 10:05 AM, Bart Schaefer wrote:

What was the reason to implement [[ ]]? What advantages it gives? As
I see [[ ]] and [ ] provides same features.

-- 
sergio.

Re: [[ and [

[Peter Stephenson]

On Tue, 6 Mar 2012 15:05:41 +0400
sergio <mailbox@sergio.spb.ru> wrote:
> On 03/06/2012 10:05 AM, Bart Schaefer wrote:
> 
> What was the reason to implement [[ ]]? What advantages it gives? As
> I see [[ ]] and [ ] provides same features.

[[ ... ]] is specially parsed: this gives better error checking and an
assurance that each argument means what you think it does.

Consider:

  foo=
  [ $foo -eq "" ]

With normal shell behaviour, this turns into the arguments "[", "-eq",
"]", which confuses the test command.  This is why you see all those
"x"s and double quotes in configure scripts.  However,

  [[ $foo -eq "" ]]

does exactly what it looks like it does because "[[" is special to the
shell and it looks at the raw arguments to parse the syntax.

Generally, I get the impression Bourne shell syntax was something of a
leap in the dark at the time; if you were designing it from scratch now,
it would look considerably different in a lot of ways.

-- 
Peter Stephenson <pws@csr.com>            Software Engineer
Tel: +44 (0)1223 692070                   Cambridge Silicon Radio Limited
Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, UK


Member of the CSR plc group of companies. CSR plc registered in England and Wales, registered number 4187346, registered office Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, United Kingdom
More information can be found at www.csr.com. Follow CSR on Twitter at http://twitter.com/CSR_PLC and read our blog at www.csr.com/blog

Re: [[ and [

[Peter Stephenson]

On Tue, 6 Mar 2012 11:35:32 +0000
Peter Stephenson <Peter.Stephenson@csr.com> wrote:
> However,
> 
>   [[ $foo -eq "" ]]
> 
> does exactly what it looks like it does

Sigh.  Sort of.  [ ... -eq ...] and [[ ... -eq ... ]] are numeric tests,
so if foo is a string of zeros it's not doing what it looks like.  I
should have said

  [ $foo = "" ]]

and

  [[ $foo = "" ]]

or

  [[ $foo == "" ]]

I don't know why it's that way round, it seems a bit illogical (which is
why Perl is the other way), but it always has been that way.

-- 
Peter Stephenson <pws@csr.com>            Software Engineer
Tel: +44 (0)1223 692070                   Cambridge Silicon Radio Limited
Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, UK


Member of the CSR plc group of companies. CSR plc registered in England and Wales, registered number 4187346, registered office Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, United Kingdom
More information can be found at www.csr.com. Follow CSR on Twitter at http://twitter.com/CSR_PLC and read our blog at www.csr.com/blog

Re: [[ and [

[Mikael Magnusson]

On 6 March 2012 12:44, Peter Stephenson <Peter.Stephenson@csr.com> wrote:
> On Tue, 6 Mar 2012 11:35:32 +0000
> Peter Stephenson <Peter.Stephenson@csr.com> wrote:
>> However,
>>
>>   [[ $foo -eq "" ]]
>>
>> does exactly what it looks like it does
>
> Sigh.  Sort of.  [ ... -eq ...] and [[ ... -eq ... ]] are numeric tests,
> so if foo is a string of zeros it's not doing what it looks like.  I
> should have said
>
>  [ $foo = "" ]]
>
> and
>
>  [[ $foo = "" ]]
>
> or
>
>  [[ $foo == "" ]]
>
> I don't know why it's that way round, it seems a bit illogical (which is
> why Perl is the other way), but it always has been that way.

Incidentally, this is probably the most confusing error message ever,
% [ $foo == "" ]
zsh: = not found

-- 
Mikael Magnusson

Re: [[ and [

[sergio]

Peter,
than you for a good explanation. Hope you'll add it to the manual.

-- 
sergio.

Re: [[ and [

[Peter Stephenson]

On Tue, 6 Mar 2012 20:16:32 +0400
sergio <mailbox@sergio.spb.ru> wrote:
> than you for a good explanation. Hope you'll add it to the manual.

Actually, the description of test and [ already has this to say...

test [ arg ... ]
[ [ arg ... ] ]
       Like the system version of test.  Added for  compatibility;  use
       conditional  expressions  instead  (see the section `Conditional
       Expressions').  The main  differences  between  the  conditional
       expression  syntax  and the test and [ builtins are:  these com‐
       mands are not handled syntactically, so  for  example  an  empty
       variable  expansion  may cause an argument to be omitted; syntax
       errors cause status 2 to be returned instead of a  shell  error;
       and  arithmetic  operators  expect integer arguments rather than
       arithmetic expressions.

       The command attempts to implement POSIX and its extensions where
       these are specified.  Unfortunately there are intrinsic ambigui‐
       ties in the  syntax;  in  particular  there  is  no  distinction
       between  test  operators  and  strings  that resemble them.  The
       standard attempts to resolve these for small  numbers  of  argu‐
       ments  (up  to  four);  for five or more arguments compatibility
       cannot be relied on.  Users are urged wherever possible  to  use
       the `[[' test syntax which does not have these ambiguities.

-- 
Peter Stephenson <pws@csr.com>            Software Engineer
Tel: +44 (0)1223 692070                   Cambridge Silicon Radio Limited
Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, UK


Member of the CSR plc group of companies. CSR plc registered in England and Wales, registered number 4187346, registered office Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, United Kingdom
More information can be found at www.csr.com. Follow CSR on Twitter at http://twitter.com/CSR_PLC and read our blog at www.csr.com/blog

Re: [[ and [

[sergio]

On 03/06/2012 08:29 PM, Peter Stephenson wrote:

> Actually, the description of test and [ already has this to say...
Sorry, my fault.

-- 
sergio.

Re: [[ and [

[Nadav Har'El]

On Tue, Mar 06, 2012, sergio wrote about "Re: [[ and [":
> On 03/06/2012 10:05 AM, Bart Schaefer wrote:
> 
> What was the reason to implement [[ ]]? What advantages it gives? As
> I see [[ ]] and [ ] provides same features.

In the original Unix, "[" was not a shell builtin - it was a normal
program, /bin/[, normally a hard link to /bin/test. As such, you needed
to follow ordinary (and annoying) quoting rules for its parameters.

When ksh (and later zsh and bash) implemented '[' as a builtin, it
retained its exact old behavior for backward compatibility, this should NOT
and instead invented a new syntax, the [[, for an improved version, with
less need for quoting and new tests, including for example the "<" operator
for text dictionary-order (on why this deals with text and not numbers,
see my separate reply).


-- 
Nadav Har'El                        |                  Wednesday, Mar 7 2012, 
nyh@math.technion.ac.il             |-----------------------------------------
Phone +972-523-790466, ICQ 13349191 |Linux: Because a PC is a terrible thing
http://nadav.harel.org.il           |to waste.

Re: [[ and [

[Nadav Har'El]

On Tue, Mar 06, 2012, Peter Stephenson wrote about "Re: [[ and [":
> On Tue, 6 Mar 2012 11:35:32 +0000
> Sigh.  Sort of.  [ ... -eq ...] and [[ ... -eq ... ]] are numeric tests,
> so if foo is a string of zeros it's not doing what it looks like.  I
> should have said
> 
>   [ $foo = "" ]]
>...
> I don't know why it's that way round, it seems a bit illogical (which is
> why Perl is the other way), but it always has been that way.

In addition to being the way it's always been, it makes even more sense
when you consider the newer features of ksh (and following it, zsh and bash).

In ksh, arithmetic expressions got a new syntax, ((. If if you would
like to check for arithmetic equality (or whatever), you'd most likely
use
	if (( $foo == 2 ))
or even better, without the "$":
	if (( foo == 2 ))
The new (( can of course do much more interesting arithmetic
expressions, e.g.:
	if (( foo+1 < bar/2.0 ))

Then, test, [, or [[ remain useful only for *strings*. So it makes sense
for their ==, <, etc., to be geared for comparing strings. Ksh even
considers "-eq", "-gt" to be obsolete - if you're using them, you should
probably be using (( instead.

The zsh manual does not state that "-eq" and the rest are obsolete (in
Conditional Expressions), but maybe it should: Ksh added this statement
to its manual in 1993, after it was absent in 1998: Compare
	http://www.research.att.com/sw/download/man/man1/ksh88.html
	http://www.research.att.com/sw/download/man/man1/ksh.html


-- 
Nadav Har'El                        |                  Wednesday, Mar 7 2012, 
nyh@math.technion.ac.il             |-----------------------------------------
Phone +972-523-790466, ICQ 13349191 |Live as if you were to die tomorrow,
http://nadav.harel.org.il           |learn as if you were to live forever.

Re: [[ and [

[Nadav Har'El]

On Tue, Mar 06, 2012, Mikael Magnusson wrote about "Re: [[ and [":
> Incidentally, this is probably the most confusing error message ever,
> % [ $foo == "" ]
> zsh: = not found

I see something slightly different. On zsh 4.3.15, with unset $foo, I
get:

	$ [ $foo == "" ]
	zsh: parse error: condition expected: ==

Which indeed is not very clear, but arguably not "the most confusing
error message ever" ;-)
To compare, GNU test shows this:

	$ /usr/bin/[ $foo == "" ]
	/usr/bin/[: missing argument after `'

Bash shows:

	$ [ $foo == "" ]
	bash: [: ==: unary operator expected

Ksh shows
	$ [ $foo == "" ]
	ksh: [: argument expected

I'm not sure which is clearest, probably ksh's.
If it were up to me, I'd say something like:

	$ [ $foo == "" ]
	zsh: parse error: ==: missing first argument.

-- 
Nadav Har'El                        |                  Wednesday, Mar 7 2012, 
nyh@math.technion.ac.il             |-----------------------------------------
Phone +972-523-790466, ICQ 13349191 |A language is a dialect with an army.
http://nadav.harel.org.il           |

Re: [[ and [

[Mikael Magnusson]

On 7 March 2012 10:02, Nadav Har'El <nyh@math.technion.ac.il> wrote:
> On Tue, Mar 06, 2012, Mikael Magnusson wrote about "Re: [[ and [":
>> Incidentally, this is probably the most confusing error message ever,
>> % [ $foo == "" ]
>> zsh: = not found
>
> I see something slightly different. On zsh 4.3.15, with unset $foo, I
> get:
>
>        $ [ $foo == "" ]
>        zsh: parse error: condition expected: ==
>
> Which indeed is not very clear, but arguably not "the most confusing
> error message ever" ;-)

You need to not have done unsetopt equals to see the confusing one :).

-- 
Mikael Magnusson

Re: [[ and [

[Nadav Har'El]

On Wed, Mar 07, 2012, Mikael Magnusson wrote about "Re: [[ and [":
> > On Tue, Mar 06, 2012, Mikael Magnusson wrote about "Re: [[ and [":
> >> Incidentally, this is probably the most confusing error message ever,
> >> % [ $foo == "" ]
> >> zsh: = not found
> >
> > I see something slightly different. On zsh 4.3.15, with unset $foo, I
>..
> You need to not have done unsetopt equals to see the confusing one :).

Wow, it's not just EQUALS (which I do have set, and use it all the time),
it's also the NOMATCH option which I forgot was the default. I've mentioned
this in the past that I think that NONOMATCH should be the default...
With NONOMATCH, I get:

$ echo =ls
/bin/ls
$ echo =dog
=dog
$ echo ==  
==
$ [ a == a ] && echo success
success

which is exactly what I would expect. With the (unfortunately default)
NOMATCH, I get:

$ echo =ls
/bin/ls
$ echo =dog
zsh: dog not found
$ echo ==
zsh: = not found
$ [ a == a ] && echo success
zsh: = not found

So the confusing error doesn't come from [, it comes from NOMATCH, which
(did I say that? :-)) I really don't like being the default.

One thing to note, though, is that the traditional equality operator for
test (or '[') was "=", not "==", so had you used =, you would not have
this problem in the first place...


-- 
Nadav Har'El                        |                  Wednesday, Mar 7 2012, 
nyh@math.technion.ac.il             |-----------------------------------------
Phone +972-523-790466, ICQ 13349191 |Topologist, n.: A person who cannot tell
http://nadav.harel.org.il           |a doughnut from a coffee mug.

Re: [[ and [

[Peter Stephenson]

On Wed, 7 Mar 2012 10:52:36 +0200
Nadav Har'El <nyh@math.technion.ac.il> wrote:
> The zsh manual does not state that "-eq" and the rest are obsolete (in
> Conditional Expressions), but maybe it should:

Actually, I don't consider them obsolete.  I prefer

[[ $n -gt 1 && $m == foo ]]

to the rather cluttered

(( n > 1 )) && [[ $m == foo ]]

There's no gain in us getting rid of them anyway, so declaring them
obsolete is confusing people (we've had this with similar features
there's no point in removing).

-- 
Peter Stephenson <pws@csr.com>            Software Engineer
Tel: +44 (0)1223 692070                   Cambridge Silicon Radio Limited
Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, UK


Member of the CSR plc group of companies. CSR plc registered in England and Wales, registered number 4187346, registered office Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, United Kingdom
More information can be found at www.csr.com. Follow CSR on Twitter at http://twitter.com/CSR_PLC and read our blog at www.csr.com/blog

Re: [[ and [

[Nadav Har'El]

On Wed, Mar 07, 2012, Peter Stephenson wrote about "Re: [[ and [":
> On Wed, 7 Mar 2012 10:52:36 +0200
> Nadav Har'El <nyh@math.technion.ac.il> wrote:
> > The zsh manual does not state that "-eq" and the rest are obsolete (in
> > Conditional Expressions), but maybe it should:
>...
> 
> There's no gain in us getting rid of them anyway, so declaring them
> obsolete is confusing people (we've had this with similar features
> there's no point in removing).

Well, I agree that -eq and friends will never be removed - they existed for
at least 33 years now (since 7th edition Unix), so let them stay another 30
years ;-)). But people who read the manual can be told that there is a
newer alternative, namely ((. Also, like I said, zsh's inspiration for
the '[[' command, ksh, already declared 20 years ago that these options
are obsolete.

Anyway, it was you who said these -eq et al. were confusing :-)

-- 
Nadav Har'El                        |                  Wednesday, Mar 7 2012, 
nyh@math.technion.ac.il             |-----------------------------------------
Phone +972-523-790466, ICQ 13349191 |You have the right to remain silent.
http://nadav.harel.org.il           |Anything you say will be used against you.

Re: [[ and [

[Peter Stephenson]

On Wed, 7 Mar 2012 16:11:22 +0200
Nadav Har'El <nyh@math.technion.ac.il> wrote:
> Well, I agree that -eq and friends will never be removed - they existed for
> at least 33 years now (since 7th edition Unix), so let them stay another 30
> years ;-)). But people who read the manual can be told that there is a
> newer alternative, namely ((.

That's true.

> Anyway, it was you who said these -eq et al. were confusing :-)

I like confusing myself, it's one of the ways I know both I and the
world still exist.

Index: Doc/Zsh/cond.yo
===================================================================
RCS file: /cvsroot/zsh/zsh/Doc/Zsh/cond.yo,v
retrieving revision 1.10
diff -p -u -r1.10 cond.yo
--- Doc/Zsh/cond.yo	24 Oct 2011 15:35:14 -0000	1.10
+++ Doc/Zsh/cond.yo	7 Mar 2012 14:24:09 -0000
@@ -151,6 +151,11 @@ based on ASCII value of their characters
 )
 item(var(exp1) tt(-eq) var(exp2))(
 true if var(exp1) is numerically equal to var(exp2).
+Note that for purely numeric comparisons use of the
+tt(LPAR()LPAR())var(...)tt(RPAR()RPAR()) builtin described in
+ifzman(the section `ARITHMETIC EVALUATION')\
+ifnzman(noderef(Arithmetic Evaluation)) is more convenient than
+conditional expressions.
 )
 item(var(exp1) tt(-ne) var(exp2))(
 true if var(exp1) is numerically not equal to var(exp2).

-- 
Peter Stephenson <pws@csr.com>            Software Engineer
Tel: +44 (0)1223 692070                   Cambridge Silicon Radio Limited
Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, UK


Member of the CSR plc group of companies. CSR plc registered in England and Wales, registered number 4187346, registered office Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, United Kingdom
More information can be found at www.csr.com. Follow CSR on Twitter at http://twitter.com/CSR_PLC and read our blog at www.csr.com/blog

Re: [[ and [

[Bart Schaefer]

On Mar 7, 11:21am, Nadav Har'El wrote:
}
} Wow, it's not just EQUALS (which I do have set, and use it all the time),
} it's also the NOMATCH option which I forgot was the default. I've mentioned
} this in the past that I think that NONOMATCH should be the default...

Everybody has different preferences; there's no default we can choose
about which someone wouldn't complain.  NO_MATCH is the default for
the simple reason that Zsh was designed to act like csh at the prompt
for people who preferred Bourne shell's more cleanly defined syntax,
and nomatch is the default in csh.

Re: [[ and [

[Mark]

>  But people who read the manual can be told that there is a
> newer alternative, namely ((.

They should also be very aware that (( 0 )) returns false and
(( "not 0" )) returns true. Try it:

(( 0 )) && echo foobar