MIME-Version: 1.0
References: <20120202102957.GB14566@csmvddesktop>
            <4F2A68F9.1010508@morningstar2.co.uk>
            <20120202111808.GA15304@csmvddesktop>
            <CANQYN6x2P2Mr+1LC5gCeUPHxTxRzmfVfPep8-oJ+VnW=5pvf1Q@mail.gmail.com>
            <4F2AC338.2060505@residenset.net>
Content-Type: text/plain; charset=ISO-8859-1
Message-ID:  <CANQYN6yb280A0c=EW70GtRH6i16fWEK2s2_C1xhyUCFnn0NOgA@mail.gmail.com>
Date:         Thu, 2 Feb 2012 13:35:58 -0500
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: Bruno Le Floch <blflatex@GMAIL.COM>
Subject: Re: Mapping Functions Versions for All and Some
To: LATEX-L@listserv.uni-heidelberg.de
In-Reply-To:  <4F2AC338.2060505@residenset.net>
Precedence: list
Status: R

Hi Lars,

>> I've been wondering: should it be called "regex" or "regexp"?
>
> My preference is for "regexp", but I can see it might by now be a bit
> awkward to change it.

My original argument in favor of regex (besides the fact that
Wikipedia's regex(p|) page uses that and that it's shorter) is that it
matches nicely with (La|)TeX. Before I planned to include it in
LaTeX3, I thought of calling the package ReGeX. :-)

>> Like Joseph, I don't see how useful the functions you propose are. Do
>> you have a precise use case?
>
> Imprecisely, I suspect one might find such predicates useful when about to
> choose between two methods of doing something: one which is quick, but only
> works in the (common) simple case, and one which is general but slow.
> Whether that happens frequently enough for them to be put in the kernel is
> another matter.

The approach I described with \tl_map_inline:nn and \prg_map_break:
seems pretty much as fast as a hard-coded version. Alternatively, for
an expandable version, you could use

    \cs_new:Npn \your_test:n #1
      { \str_if_eq:nnT { bingo! } {#1} { \use_ii:nnn } }
    %....
    \tl_map_function:NN \l_your_tl \your_test \use_ii:nn
       { True code }
       { False code }

Of course, the non-expandable version is cleaner. But I somewhat doubt
that people who really need the expandable version can't code that
easily. Dunno.

> sidetracked by other projects (in this case, actual research)

Research does that to people :-). I'm off to work on my PhD for a short bit.

Cheers,
Bruno