Re: [xml-dev] version numbers and infosets

At 2:47 PM -0400 7/23/02, John Cowan wrote:


>People use mainframes, and XML too.  If XML 1.0 had insisted
>that only CR-LF and LF were acceptable line terminators, don't you
>think an argument based on justice for Mac users would have been
>appropriate?

No, I don't. If XML 1.0 had not allowed bare carriage returns, then I 
would expect Mac users to adjust. Guess what? They could. Today, I 
easily and routinely process and generate XML ending in CR/LF and LF 
only on my Mac. No hassles. Do you really think mainframe programmers 
are so much stupider than Mac users that they can't do this too? Or 
that a multi-million dollar mainframe is that much less capable than 
a $1500 iMac?

The time to make this argument was in 1996 when XML 1.0 was being 
developed. If at that point, bare CRs had been ruled out, I might 
have argued for them. If at that point, you had argued that NEL 
should be added to the white space production, I might have supported 
you. That time has past. The costs of the change now vastly outweigh 
the benefits.

Let's try to quantify those benefits. How many people use mainframes 
today in a way that XML 1.0 presents them with problems? So far I 
have yet to encounter *ONE* person who is actually inconvenienced by 
XML in this fashion.

I challenge the XML working group to present a case based on actual 
software and hardware used today in which the lack of NEL is a 
problem. Specifically,

What currently sold, shipped, or supported software can handle NEL 
but not CR or CR/LF? How many users does this software have, and how 
many of these users actually need to use this software to process 
XML? (Airline ticket agents typing forms into dumb terminals or any 
other user who never actually sees or thinks about the XML doesn't 
count.) Feel free to list more than one.

Let us also stop pretending that there are no costs associated with 
releasing a new version of XML into the world. Let us tally the costs 
of all the users who will upgrade and compare that to the benefit to 
be achieved by releasing XML 1.1.

If you persist in this ridiculous claim that NEL is a matter of 
justice, then let us consider the justice of forcing many users 
around the world to upgrade their systems to support the few large 
businesses and computer companies that still use mainframes. Is it 
more just that the cost be borne be IBM and Citigroup or by tens of 
thousands of individual developers?

>You mention the legal principle of "stare decisis".  This is by no means
>applied in every area of the law, and in particular gives way before
>claims of natural equity.  It is *not* always more important that the
>law be unchanging than that it do justice or right.

In a question of justice or rightness, no it is not. However, this is 
simply not a question of justice in any plausible moral system I have 
ever encountered. I do not understand how you can possibly phrase the 
debate this in those terms. It is a question of costs and benefits. 
There is more than one way to do it, and while some ways may be 
better than others, the benefits of sticking with the established 
way, even if it less than optimal, often outweigh the costs of 
switching.
-- 

+-----------------------+------------------------+-------------------+
| Elliotte Rusty Harold | elharo@metalab.unc.edu | Writer/Programmer |
+-----------------------+------------------------+-------------------+
|          XML in a  Nutshell, 2nd Edition (O'Reilly, 2002)          |
|              http://www.cafeconleche.org/books/xian2/              |
|  http://www.amazon.com/exec/obidos/ISBN%3D0596002920/cafeaulaitA/  |
+----------------------------------+---------------------------------+
|  Read Cafe au Lait for Java News:  http://www.cafeaulait.org/      |
|  Read Cafe con Leche for XML News: http://www.cafeconleche.org/    |
+----------------------------------+---------------------------------+

Received on Tuesday, 23 July 2002 15:34:18 UTC