Mime-Version: 1.0
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
Message-ID:  <LISTSERV%200810012115133110.042A@LISTSERV.UNI-HEIDELBERG.DE>
Date:         Wed, 1 Oct 2008 21:15:13 +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: Joseph Wright <joseph.wright@MORNINGSTAR2.CO.UK>
Subject: Re: Something like \@removeelement
To: LATEX-L@LISTSERV.UNI-HEIDELBERG.DE
Precedence: list
Status: R

Frank Mittelbach wrote:
> Joseph Wright writes:
>  > Hi all,
>  > 
>  > On my latest LaTeX3 excursion, I'm looking achieve the effect of
>  > \@removeelement. I'd expect this to be a clist function, but there is
>  > nothing like "\clist_remove_elt:nnN" (#1 = elt, #2 = clist, #3 = new
>  > clist name) or "\clist_remove_elt:nN" (#1 = elt, #2 = clist to be edited).
> 
> what are the precise semantics that you have in mind (with respect to element
> not there, element there more than once etc ...) ?

Good questions.  In my case, I've already checked that the element is
present and I can use \clist_remove_duplicates:N, so I'm not too fussed. 
However, in general this needs to be addressed.

I'd imagine that, in the main, any removal function should tolerate the
element not being there at all.  \clist_if_in:NnTF is already available, and
so can be used for checks by programmers.  I'd imagine that most uses where
an element is to be removed will either have a check before removal or will
not need to know if the element is not there.  

On the repeated-item question, I think that in general you can't rely on
\clist_remove_duplicates:N.  So perhaps two versions are needed:

\clist_remove_elt_first:Nn (or perhaps _left:)
\clist_remove_elt_all:Nn   (or perhaps _total: or _complete:)

I'm thinking of usage along the lines:

\clist_set:Nn \l_example_clist {a,b,c,d,d,e,e}

\clist_remove_elt_first:Nn \l_example_clist {a}
% \l_example_clist now {b,c,d,d,e,e}
\clist_remove_elt_first:Nn \l_example_clist {d}
% \l_example_clist now {b,c,d,e,e}
\clist_remove_elt_all:Nn \l_example_clist {e}
% \l_example_clist now {b,c,d}

Here, I'm assuming that the clist is supplied by name and is returned into
the same list. \@removeelement actually works slightly differently,  but I
think that the syntax above is clearer.  I'm not sure a pre-expanded clist
is very helpful, so I'd probably avoid:

\clist_remove_elt:nnN {a,b,c} {b} \l_new_clist

(which is actually similar to \@removeelement}.  On the other hand, it might
be useful to store the new clist in a separate place to the initial one:

\clist_set:Nn \l_example_clist {a,b,c,d,d,e,e}
\clist_set:Nn \l_new_clist {}

\clist_remove_elt_first:NnN \l_example_clist {a} \l_new_clist
% \l_new_clist now {b,c,d,d,e,e}
\clist_remove_elt_first:NnN \l_example_clist {d} \l_new_clist
% \l_new_clist now {a,b,c,d,e,e}
\clist_remove_elt_all:NnN \l_example_clist {e} \l_new_clist
% \l_new_clist now {a,b,c,d,d}

How does all of this sound?
--
Joseph Wright