User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:24.0) Gecko/20100101
            Thunderbird/24.5.0
MIME-Version: 1.0
References: <5367BB2E.1010403@nag.co.uk> <j39taff9cl0n.dlg@nililand.de>
Content-Type: multipart/alternative; boundary="------------000604010607020108070609"
Received-SPF: None (: nag.co.uk does not designate permitted sender hosts)
Message-ID:  <536B472F.1000002@nag.co.uk>
Date:         Thu, 8 May 2014 09:58:23 +0100
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: David Carlisle <davidc@NAG.CO.UK>
Organization: NAG
Subject: Re: New LaTeX2e Release 2014/05/01 : Serious bug in keyval?
To: LATEX-L@LISTSERV.UNI-HEIDELBERG.DE
In-Reply-To:  <j39taff9cl0n.dlg@nililand.de>
Precedence: list
Envelope-To: <rainer.schoepf@GMX.NET>
Status: R

--------------000604010607020108070609
Content-Type: text/plain; charset="ISO-8859-1"; format=flowed
Content-Transfer-Encoding: 7bit

On 08/05/2014 09:14, Ulrike Fischer wrote:
> Comparing an output with  \tracingmacros one can see that the
 > behaviour of keyval has changed.

This is of course the problem with fixing bugs after a long time.
But if we refuse to fix _anything_ then not everyone will be happy either.

It's a documented change in ltnews21 and the changes file,
although that doesn't necessarily mean it shouldn't be backed out
if it is wrong, or if it is right in an ideal world but not the messy
world with legacy issues that we live in.


The intended behaviour of just dropping one brace group is more sensible

a={{{b}}}

strips one {} needed for hiding commas and leaves the value as {{b}}

The old version always stripped two brace groups so the value would have 
been {b}

Replying now having just seen your mail, haven't checked the test case yet.

It's possible there is a bug in the new implementation and it's doing 
the wrong thing
in which case it's obvious what to do: fix it.

Or it may be the new implementation is working as designed but breaking 
existing code.
Then have to take the decision to push for the other package to update, 
or take the
decision to roll back this change.

looking....

David


-- 
google plus: https:/profiles.google.com/d.p.carlisle


________________________________________________________________________
The Numerical Algorithms Group Ltd is a company registered in England
and Wales with company number 1249803. The registered office is:
Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.

This e-mail has been scanned for all viruses by Star. The service is
powered by MessageLabs. 
________________________________________________________________________
--------------000604010607020108070609
Content-Type: text/html; charset="ISO-8859-1"
Content-Transfer-Encoding: 7bit

<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    On 08/05/2014 09:14, Ulrike Fischer wrote:<br>
    <span style="white-space: pre;">&gt; Comparing an output with
      \tracingmacros one can see that the <br>
      &gt; behaviour of keyval has changed.</span><br>
    <br>
    This is of course the problem with fixing bugs after a long time.<br>
    But if we refuse to fix _anything_ then not everyone will be happy
    either.<br>
    <br>
    It's a documented change in ltnews21 and the changes file,<br>
    although that doesn't necessarily mean it shouldn't be backed out<br>
    if it is wrong, or if it is right in an ideal world but not the
    messy<br>
    world with legacy issues that we live in.<br>
    <br>
    <br>
    The intended behaviour of just dropping one brace group is more
    sensible<br>
    <br>
    a={{{b}}}<br>
    <br>
    strips one {} needed for hiding commas and leaves the value as {{b}}<br>
    <br>
    The old version always stripped two brace groups so the value would
    have been {b}<br>
    <br>
    Replying now having just seen your mail, haven't checked the test
    case yet.<br>
    <br>
    It's possible there is a bug in the new implementation and it's
    doing the wrong thing<br>
    in which case it's obvious what to do: fix it.<br>
    <br>
    Or it may be the new implementation is working as designed but
    breaking existing code.<br>
    Then have to take the decision to push for the other package to
    update, or take the<br>
    decision to roll back this change.<br>
    <br>
    looking....<br>
    <br>
    David<br>
    <br>
    <br>
    <br>
    <br>
    <br>
    <br>
    <br>
    <pre class="moz-signature" cols="72">-- 
google plus: <a class="moz-txt-link-freetext" href="https:/profiles.google.com/d.p.carlisle">https:/profiles.google.com/d.p.carlisle</a></pre>
  <br clear="both">
________________________________________________________________________<BR>
The Numerical Algorithms Group Ltd is a company registered in England<BR>
and Wales with company number 1249803. The registered office is:<BR>
Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.<BR>
<BR>
This e-mail has been scanned for all viruses by Star. The service is<BR>
powered by MessageLabs. <BR>
________________________________________________________________________<BR>
</body>
</html>


--------------000604010607020108070609--