Re: [xml-dev] Why would MS want to make XML break on UNIX, Perl, Python, etc ?

From
Rick Jelliffe <>
To
Date
2001-12-21T08:38:47Z
ID
<033901c189fa$e8775ad0$>
Thread
Re: [xml-dev] Why would MS want to make XML break on UNIX, Perl, Python, etc ?
From: "Richard Tobin" <>
 
> >>> So if you can send ^D down your UNIX pipes, it is because the
> >>> specific programs in the
> >>> pipeline happen to be reading the input stream in binary mode not =
> >>> text mode.
> 
> But this is not true.  It would only not work if some specific program
> had its own "text mode" in which it did this.  I have never seen such
> a program.
> 
> Do you have a real example of one?

From shell implementation CYG-WIN (which is admittedly not UNIX)
 http://www.cygwin.com/faq/faq.txt
   "There's a new registry variable "fmode=binary" which controls
whether the tools always open files in binary mode (unless overridden
with O_TEXT), or always open files in text mode (unless overridden with
O_BINARY)."
 
For more on this, see 
http://sources.redhat.com/cygwin/cygwin-ug-net/using-textbinary.html

See also the discussion of booking binary files out of CVS at
http://www.cygwin.com/ml/cygwin-apps/2001-04/threads.html#00081
at "ash stdin mode is set to O_TEXT, "

On PCs, the DOS command "type" stops on ^Z when writing to the console. 
If you use "type x.txt | more" it does not truncate. (Its help specifically says
it is for "text".) 

The DOS command "copy" stops on ^Z if you use /A option. 

But I have been confused with talking of pipes and stdio when the APIs I 
mentioned all deal with file IO.  The mechanics I gave were wrong, however 
the problem is still real. 

Cheers
Rick Jelliffe