L2/99-324

Mapping tables index announcement and comments
From: Kent Karlsson and Ken Whistler

1999-10-12


Ken Whistlers comments to Kent's text below

Kent,

Lisa and Arnold are putting this on the agenda of the upcoming 
UTC meeting to discuss.

Here are some early answers to some of your questions.

1. MS and Apple may not agree on their Mac mappings. That is an
   issue those two entities need to work out between themselves.

2. I believe the MS EBCDIC tables largely follow the IBM
   CDRA mappings, but the table formats they use are distinct.

3. IBM made a policy decision not to post its mapping tables
   on the Unicode website. The CD they are talking about is
   the CD that comes with the IBM publication: Character Data
   Representation Architecture, Reference and Registry. That
   has software on it that you need to install and run in order
   to extract mapping information. I have already tweaked IBM
   about that -- they should at *least* put plain text mapping
   tables on the CD, in addition to the API-driven extraction.

4. We are aware of the problem with the Apple tables using CR
   as EOL. I am hoping we can fix that before freezing the MAPPINGS
   directory for our CD-ROM for Unicode 3.0.

5. I am aware that some of the tables end in a ^Z. I'm not sure
   we'll fix that, however, since it is a less critical problem.

6. The mappings for C0 controls are policy to add, for those
   tables that are under our control, but the vendors own
   their own tables, and we cannot force them to construct
   their tables in a certain way.

--Ken


Kent Karlsson's contribution

Hi! 
        Perhaps the attached file (an index/readme file for the 
Unicode MAPPINGS directory) could be reviewed at the next UTC 
meeting?  (Ken and Rick have seen a slightly earlier version, 
*without* the perhaps not-all-too-neutral "obsolecsent" remarks.) 

        Two things surprise(d) me regarding the MAPPINGS directory: 
1) MS has a few mapping files for Mac and EBCDIC. The MS Mac 
   ones differ slightly from the corresponding Apple Mac ones. 
   Whether the MS EBCDIC ones differ from the IBM EBCDIC ones, 
   I cannot determine due to point 2. 
2) IBM has none of its mapping files posted, instead there 
   is just a reference to an IBM CD with "binary" (whatever 
   that means) mapping data. 

There are also some slight problems with some of the mapping 
files themselves: 

a) some use CR as line ending, though most use LF. 
b) some end in ctrl-Z, which they should not. 
c) some include mappings for C0 controls, some don't (even if 
   they use C0 control 'characters' at least to some extent). 
d) comment conventions vary slightly (which makes "diff" harder 
   since one need to 'normalise' the comments before 'diff'-ing. 
e) Adobe uses their own format for no apparent reason. 

And as for ftp: just clicking on an ftp link, then it takes a 
long time before one gets the file (anonymous login delays). 
If they were made available by http (too), that would be quicker 
to use. 

Comment 0: The references to the UTRs should be made into hyperlinks. 

I will not be attending the UTC meeting, so either you have 
to email me any comments/requests for changes, or someone else 
will have to assume the future editing of this file (if you 
at all want to have this file). 

                Kind regards 
                /kent k
Page 2		C:\WINNT\Profiles\winkleaf\Application Data\Microsoft\Templates\Normal.dot