References: <CACKZBPWfcegW+5axif=KqKc_cAzpRU5dZq2U4Ti9q8A+zaq+FA@mail.gmail.com>
            <CANQYN6x3Hp5L0NoxnfEXF8Uk1GGFTOuZajeeuGCJpRSHp8HREQ@mail.gmail.com>
            <4EF193C0-A612-4C83-B39A-35BAEDC305C6@mittelbach-online.de>
Mime-Version: 1.0 (1.0)
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 8bit
Message-ID:  <60AE4AED-1FBB-4159-9F64-52A7936EC806@mittelbach-online.de>
Date:         Tue, 16 Jul 2013 21:13:22 +0200
Reply-To: Mailing list for the LaTeX3 project
              <LATEX-L@LISTSERV.UNI-HEIDELBERG.DE>
Sender: Mailing list for the LaTeX3 project <LATEX-L@LISTSERV.UNI-HEIDELBERG.DE>
From: iPad Mittelbach <ipad@MITTELBACH-ONLINE.DE>
Subject: Re: Request for argument specifiers which generate unique csnames
To: LATEX-L@LISTSERV.UNI-HEIDELBERG.DE
In-Reply-To:  <4EF193C0-A612-4C83-B39A-35BAEDC305C6@mittelbach-online.de>
Precedence: list
Envelope-To: <rainer.schoepf@GMX.NET>
Status: R

Account was locked so retry

> I agree with a) that having a mechanism for generating unique cs names is useful and b) that extending the arg specs for this is not.
> 
> In my opinion the correct approach is to return the new cs name in a tl variable from which it then can be used.
> 
> After all, to be useful you need to be able not just to generate such a cs name but also to reuse it in other places using the name that was generated before.
> 
> That allows, for example, to generate pointer data structures, where  each data record is stored in some cs name (unique) and links are made up by pointing to other records from fields in the record data structure.
> 
> I used that to build out the data structure for the LDB, e.g., you will find some code for this already in the file called l3precom.dtx (I think, can't check here with just an iPad available)
> 
> I'm not saying that this particular implementation is the way it should be, but something along the line is how I think the direction should/could be
> 
> Frank
> 
> ... written on the iPad