Re: requirements for anonymous dimensions (and question for John)

ed@xxxxxxxxxxxxxxxx wrote:

From our requirements on anonymous dimensions:

* Dimensions do not always require a name.
By this do we mearly mean that I can accept a dimension with a zero
length name, and that's it?

Otherwise it acts exactly like a regular dimension?

If so, this will be an easy one!

John, was there something else you had in mind for this that I am missing?
Ed
anonymous dimensions are local to the variable, as opposed to globally scoped dimensions which can be shared across variables. anonymous dimensions can be nameless, but it would be ok if they had a name if someone had an idea why that would be useful.

From owner-netcdf-hdf@xxxxxxxxxxxxxxxx 18 2003 Dec -0700 15:27:42
Message-ID: <wrxiskd3go1.fsf@xxxxxxxxxxxxxxxxxxxxxxx>
Date: 18 Dec 2003 15:27:42 -0700
From: Ed Hartnett <ed@xxxxxxxxxxxxxxxx>
In-Reply-To: <3FE21450.4060507@xxxxxxxxxxxxxxxx>
To: netcdf-hdf@xxxxxxxxxxxxxxxx
Subject: Re: requirements for anonymous dimensions (and question for John)
Received: (from majordo@localhost)
        by unidata.ucar.edu (UCAR/Unidata) id hBIMRhFS014614
        for netcdf-hdf-out; Thu, 18 Dec 2003 15:27:43 -0700 (MST)
Received: from rodney.unidata.ucar.edu (rodney.unidata.ucar.edu 
[128.117.140.88])
        by unidata.ucar.edu (UCAR/Unidata) with ESMTP id hBIMRgp2014610
        for <netcdf-hdf@xxxxxxxxxxxxxxxx>; Thu, 18 Dec 2003 15:27:43 -0700 (MST)
Organization: UCAR/Unidata
Keywords: 200312182227.hBIMRgp2014610
References: <u65ge2hvh.fsf@xxxxxxxxxx> <3FE21450.4060507@xxxxxxxxxxxxxxxx>
Lines: 18
User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.2
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Sender: owner-netcdf-hdf@xxxxxxxxxxxxxxxx
Precedence: bulk

John Caron <caron@xxxxxxxxxxxxxxxx> writes:

anonymous dimensions are local to the variable, as opposed to globally
scoped dimensions which can be shared across variables. anonymous
dimensions can be nameless, but it would be ok if they had a name if
someone had an idea why that would be useful.

Then how do their ids work?

If I define three shared dimensions, and get ids 0, 1, 2, then I
define an anonymous dimension, and if gets id 4, then I define another
shared dimension and it gets id 5, right?

So what does it gain to have it be local to the variable? I still have
to reserve the id, unless we want to change the nc_def_var to include
the anonymous dimensions.

Ed