Re: [Help-bash] Order of element in associative array expansion

[Davide Brini]

On Thu, 17 Oct 2013 15:37:09 -0400, Greg Wooledge <address@hidden>
wrote:

> > On 10/17/13 12:23 PM, Davide Brini wrote:
> > > I know that the order of keys and values obtained by expanding an
> > > associative array with "address@hidden" and "address@hidden" respectively 
> > > is
> > > unpredictable, which is fine.
> > > 
> > > However my question is: at a given point in time, without changes or
> > > other operations happening in between, are the two expansions
> > > guaranteed to return the keys in the same order as the respective
> > > values and viceversa?
> 
> I can't resist asking: why does it matter?  Under what circumstances
> would you traverse these two lists independently, but wish to have them
> match up properly?
> 
> Obfuscated programming contest?

Not really. A question from a friend got me thinking...

It could be a simple-minded way to serialize a bash hash and pass it to
another program or script. Simple example (pseudocode):

someotherscript "address@hidden" "address@hidden"

and in the other script you could do

for (i = 1; i <= argc / 2; i++)
  hash[argv[i]] = argv[i + argc / 2 ]

and at the end you have rebuilt the hash inside the other program. However,
this requires that the expansion of the keys and the values happen in the
same order, which is what I asked.

I'm not saying this is an especially good/optimal/whatever way to do
things and I'm perfectly aware a zillion of other ways exist, so let's
leave this aside; I've just striclty answered the question.

-- 
D.