Mime-Version: 1.0
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
Message-ID:  <LISTSERV%200810200930180505.0142@LISTSERV.UNI-HEIDELBERG.DE>
Date:         Mon, 20 Oct 2008 09:30:18 +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: Tools for processing numbers
To: LATEX-L@LISTSERV.UNI-HEIDELBERG.DE
Precedence: list
Status: R

On Tue, 14 Oct 2008 13:04:20 +0100, Joseph Wright
<joseph.wright@MORNINGSTAR2.CO.UK> wrote:

>Will Robertson wrote:
>> These ideas do sound good to me.
>> Especially reliable detection methods like \ifinteger and so on.
>
>I need to see how this is done in etoolbox, as expansion could be an
>issue.  I'll see what I can do (while retaining generality).

I suspect this will have to be non-expandable, even using the tricks from
biblatex.  What general tests do others think would be useful; at present,
I'm planning (with variants):

\numtest_if_dec:nTF % A decimal integer 
\numtest_if_bin:nTF  % Binary integer
\numtest_if_oct:nTF % Octal integer
\numtest_if_hex:nTF % Hexidecimal integer

\numtest_if_numeral:nTF % A numeral, so 0-9 plus roman numeral characters
\numtest_if_range:nTF    % A range of numbers, so in general any "numeral" plus
                                  % a selection of range characters

This is essentially the tests from biblatex generalised a little.  The range
one will not be rigorous (content as a whole checked, not individual
characters).  However, I think that the second part of the idea (general
number parsing) will cover that if more slowly.
--
Joseph Wright