Re: Netscape 4.0 press release at their server

At 10:35a -0500 10/18/96, Scott E. Preece wrote:
>Nothing is wrong with UNIX magic numbers as an implementation of file
>typing, except that they are weakly integrated - they don't exist for
>many kinds of files, there is no registry or namespace allocation
>convention (actually, there is, or was, though I don't know how widely
>known it is or who is doing it now that AT&T no longer owns UNIX), there
>is no way to manipulate them, etc.

Manipulating them is easy enough, though it would touch the modification
date of the file (ick) and could easily screw up the file... ;)
I'm actually in the process of designing a standard file header for all my
programming projects' support files, and would be interested in any specs
on those magic number thingies...

At 11:49a -0700 10/18/96, David Perrell wrote:
>Walter Ian Kaye wrote:
>> The difference is that Apple *registers* creator type (parent
>> application) and file type codes. This prevents Word/FrameMaker and
>> similar conflicts. Of course, another difference is that the 4-byte
>> codes are  *binary*, resulting in 4 billion possible codes. I've even
>> reserved a few for my own MacOS programs! ;-)
>With what central authority must you register these codes

<DT><A HREF=""
ADD_DATE="833590662" LAST_VISIT="835996893"
LAST_MODIFIED="833590504">Creator/File Type Registration Requests</A>

(copied/pasted from my Bookmarks.html file)

>and how does this facilitate interoperability with other systems?

Umm... it's standardized? You'll know that a file type of 'PNTG' is a
MacPaint file, regardless of what application created it. There's a FAQ
there too, though it's kinda short. Maybe there should be a SAQ, too... LOL

    Walter Ian Kaye <>     Programmer - Excel, AppleScript,
          Mountain View, CA                         ProTERM, FoxPro, HTML     Musician - Guitarist, Songwriter

Received on Friday, 18 October 1996 20:11:43 UTC