🚀 go-pugleaf

Submitted-by: david@mks.com (David Rowley)

In article <16p6bmINNs1l@ftp.UU.NET> mskuhn@immd4.informatik.uni-erlangen.de (Markus Kuhn) writes:
>But how may UCS-4 files be identified? Do they always begin with 0000feff
>and are converted if they begin with fffe0000 or other permutations?
>Does ISO 10646 say anything about this or will any future POSIX extension do?

Being relatively new to ISO 10646, I believe the intent is to
use the UCS Transformation Format (Annex F of 10646) as the
standard external representation format (such as file contents,
etc.).  This multibyte encoding supports both the UCS2 and UCS4
codeplanes.

Note that UTF and 8-bit Latin 1 (ISO 8859-1) are identical for
characters 0x00 to 0x9f.  Codepoints above 0x9f are used to
introduce the multibyte sequences.

One problem, though, is that the UTF description in ISO 10646 is
informative, rather than normative.  With this being the case
can implementors safely point to UTF as a standard encoding?

-- 
David Rowley
Mortice Kern Systems Inc.
35 King Street North, Waterloo, ON, Canada N2J 2W9
519/884-2251, FAX 519/884-8861, david@mks.com


Volume-Number: Volume 29, Number 3

Submitted-by: mohta@necom830.cc.titech.ac.jp (Masataka Ohta)

In article <16p6bmINNs1l@ftp.UU.NET>
	mskuhn@immd4.informatik.uni-erlangen.de (Markus Kuhn) writes:

>How UCS-2 files have to be handeled under future OS versions (e.g. UNIX)
>seems to be quite obvious:
>
>  - Every UCS-2 file begins with feff. If it begins with fffe, than library
>    routines will activate a 'byte order swap mode' that corrects the
>    data from an otherendian machine.

What?

How can 'cat' know the file being read is a text file?

Do you want to introduce an infamous "FILE TYPE" to UNIX?

>  - In this way, every UNIX tool (cc, cat, ...) can easily determine,
>    how the file has to be interpreted, because everything starting
>    with something else is considered to be an 8-bit Latin 1 encoded
>    file (if it is interpreted as a 'text file' at all).

What if a 8-bit Latin 1 file begins with 0xfffe?

Code points 0xfe and 0xff represent valid Latin 1 characters.

	0xfe: LATIN SMALL LETTER THORN
	0xff: LATIN SMALL LETTER Y WITH DIAERESIS

Can you still say "quite obvious"?

>But how may UCS-4 files be identified? Do they always begin with 0000feff
>and are converted if they begin with fffe0000 or other permutations?
>Does ISO 10646 say anything about this or will any future POSIX extension do?

It is one of the well known defects of ISO 10646, which the standardizing
committee simply neglected.

							Masataka Ohta


Volume-Number: Volume 29, Number 5

Submitted-by: peter@ferranti.com (Peter da Silva)

In article <16rpgaINNol0@ftp.UU.NET> david@mks.com (David Rowley) writes:
> Note that UTF and 8-bit Latin 1 (ISO 8859-1) are identical for
> characters 0x00 to 0x9f.  Codepoints above 0x9f are used to
> introduce the multibyte sequences.

That seems strange. 0x80 through 0x9f are all controls, and all the
national characters in Latin-1 are in 0xA0 to 0xFF. Why would they allow
Latin-1 control codes (CSI, etc) and blow off all the graphics? Are you
sure they didn't overload the high control range (0x80 to 0x9f)? That
would seem a much more useful encoding.
-- 
Peter da Silva
Ferranti Intl. Ctls. Corp.      Sugar Land, TX  77487-5012       +1 713 274 5180


Volume-Number: Volume 29, Number 8

Submitted-by: brnstnd@KRAMDEN.ACF.NYU.EDU (D. J. Bernstein)

Markus Kuhn writes:
>   - In this way, every UNIX tool (cc, cat, ...) can easily determine,
>     how the file has to be interpreted,

Interfaces should be _small_. By definition a small interface has simple
semantics. Simple semantics leave no room for bugs. A small interface is
programmer-friendly.

Version 7 UNIX had small interfaces. The kernel's system call interface
was small: just a few dozen calls with semantics defined in a few pages.
Each utility's external interface was small: there were almost no
conventions weighing things down. Even the internal interfaces were
small: most tools used mere pages of code.

File types infect practically every program on the system. Witness VMS.
Every program's semantics has to include its treatment of file types.
This means that the total interface size of the system becomes much,
much, much larger.

Need I say more?

Followups to alt.religion.computers.

---Dan

[ I did set Followups to alt.religion.computers, but only because Dan
  wanted it. -- mod ]

Volume-Number: Volume 29, Number 9