Re: local/typeset stupidity

[Peter Stephenson <pws@csr.com>]

Nikolai Weibull wrote:
> Why the <profanity> doesn't local work like one would expect?  If i
> write
> 
> func () {
> 	local s="..."
> }
> 
> then I don't expect zsh to respond with
> 
> func:local:1: not valid in this context: 100
> 
> right?

There's something you're not telling us.  With the default zsh options
the syntax you show is guaranteed to work.  My guess is something is
causing what you show in "..." to be split into words.  Is there a "$@"
inside it?  For example,

func() {
   local s="$@"
}
func foo 100

would produce an error like the one you show, because the line is
effectively split up as

local s="foo" 100

To assign all the arguments to s without splitting you would need
s="$*".

> local s; s="..."
> 
> works fine, though, even with typesetsilent off.

s="..."

is special syntax; it's recognised as a scalar assignment, so
wordsplitting is implicitly turned off.  This differs from "local" which
has the semantics of a normal builtin, almost (actually, there's special
behaviour of ~'s and ='s at the start of the value).

This is rather nasty.  People often expect assignments after "local"
etc. to be the same as assignments on their own, but that syntax
conflicts with the way the arguments of builtins, and any other
commands, are handled.  There's no simple answer, but retaining builtin
syntax is at least predictable and stops the already horrendous code for
local variables from becoming any worse.  Any way round is a kludge of
some sort.

>  Even with
> 
> setopt typesetsilent # (how can't this be the default?)

typesetsilent works around a different problem:

func() {
  local s;
  local s;
}
func

outputs

s=''

unless the option is turned on.  This came from the very early days of
zsh.  It was mostly useful for examining the state of variables that
already exist interactively; it's not obvious you would use it in a
function.  It's annoying this is the default behaviour.  There
may be a better way of suppressing the output in functions, but I'm not
sure whose functions it would break.  Personally I'd be quite happy to
have the effect of typesetsilent enforced inside functions.  A
compromise would be to keep the non-silent behaviour only for "typeset"
itself.

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


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.

www.mimesweeper.com
**********************************************************************