Possible bug with parameter expansion

Possible bug with parameter expansion

[Oliver Kiddle]

[-- Attachment #1: param --]
[-- Type: text/plain, Size: 397 bytes --]

The ${name:#pattern} form of parameter expansion doesn't work within
double quotes. It is acting as if the pattern matched all the elements.
The following should demonstrate:

% a=( a d=rt b=fg c e=yt )
% echo ${a:#*=*}
a c
% echo "${a:#*=*}"

% echo "${(M)a:#*=*}"
a d=rt b=fg c e=yt

This looks like a bug to me because I can't see why the double quotes
should be affecting the result.

Oliver

Re: Possible bug with parameter expansion

[Bart Schaefer]

On Mar 7, 11:36am, Oliver Kiddle wrote:
} Subject: Possible bug with parameter expansion
}
} The ${name:#pattern} form of parameter expansion doesn't work within
} double quotes. It is acting as if the pattern matched all the elements.

If you read Peter's 11-step "summary of the rules for substitution" you
will see that (4. *Double-Quoted Joining*) comes before (6. *Modifiers*).

} % echo "${(M)a:#*=*}"
} a d=rt b=fg c e=yt

zagzig% echo "${(@M)a:#*=*}" 
d=rt b=fg e=yt

} This looks like a bug to me because I can't see why the double quotes
} should be affecting the result.

Because they do.  I don't recall why joining happens before modifiers,
but I'm pretty sure there was a good reason for it at the time.  Does
this disagree with other shells that support ${var#pat}, e.g. ksh?

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com

Zsh: http://www.zsh.org | PHPerl Project: http://phperl.sourceforge.net   

Re: Possible bug with parameter expansion

[Oliver Kiddle]

On Thu, Mar 08, 2001 at 04:33:15PM +0000, Bart Schaefer wrote:

> If you read Peter's 11-step "summary of the rules for substitution" you
> will see that (4. *Double-Quoted Joining*) comes before (6. *Modifiers*).

fair enough. Thanks.

> Because they do.  I don't recall why joining happens before modifiers,
> but I'm pretty sure there was a good reason for it at the time.  Does
> this disagree with other shells that support ${var#pat}, e.g. ksh?

In ksh, there is no way I know of making var refer to an array because it
expands to the first element of the array and with [@], you get a bad
substitution.

Comparing with bash, I have to use [@] to get an array before the
modifier. With the [@], both bash and zsh work the same regardless of
quotes. So basically, I haven't been able to create a comparable situation
in bash or ksh.

I still think the behaviour seems odd mainly because I'm used to reading
these substitutions from the inside out and the double quotes are on the
outside.

Oliver