I found this funny

[nbastin at mac.com]

On Saturday, July 27, 2002, at 06:03 PM, Stewart Stremler wrote:
>
> Extensions determine the the 'type' of the file. They are an indication
> as to how the OS will interact with that file.

How is that any different from a .txt 'suffix' in Windows?  That 
determines how the OS will interact with the file.  In both DOS and 
Win32 you could change the extension, so doesn't that fit your 
definition of 'suffix'?

> Unix has the abstraction that "everything is a file, and a file is just
> a sequence of bytes".  So you can open a .gz file with an editor, or
> use cat to concatenate text files with binary files. The system doesn't
> care. A file is a file.

But isn't that a good idea?  I *like* being about to use cat for any 
kind of file, regardless of its' type.  Besides, if you really want to 
know the type, you should be using the magic, and not the extension.

> MSDOS still shows traces of this 'previous' thinking. In MSDOS, you
> have text files, and binary files. How a program interacts with these
> files depends on the 'type' -- the "extension".

You have yet to show me the distinction between suffix and extension.  
Call it what you will, but .txt as a suffix in windows still determines 
how the system deals with files.  Isn't than an extension, by your 
reasoning?  So, for example, does .idontknowwhatsuffixesare, as that 
will cause the system to open the 'Open With...' chooser.  Are you 
saying that the first one is an extension, but the second one is a 
suffix?  If this is a matter of OS recognition, then I'll give up right 
now, because I think an extension is an extension, even if your system 
doesn't recognize it.

>>> Extensions suck. *nix philosophy allows for suffixes, not extensions.
>>>
>>> The fact that Apache has determined that extensions are convenient 
>>> just
>>> goes to show that corruption spreads even to the hearts of the
>>> enlightened.
>>
>> Then in your book, *nothing* is an extension, and you're suggesting a
>> wholly new meaning for the word 'extension' in the computer world.
>
> Nope.
>
> It's just that "suffix" and "extension" are used as synonyms, that the
> distinction has become fuzzy.
>
> If you look at the definition at:
>
> http://www.learnthat.com/define/e/extension.shtml
>
> It says that the stuff before the . is the filename, and that the stuff
> after the . is the extension.  A moment's contemplation with the Linux
> filesystem should reveal that this doesn't quite apply -- what's the
> filename and what's the extension to "some.file-
> verstion_1.0.a.tar.gz.uue" ?

It's a gzipped tar file that's uuencoded.  You don't think you should be 
able to stack extensions?  Personally, I think the restriction that you 
can't have a '.' in a filename in win32 is stupid.  It thinks everything 
after that point is an extension.  Now you're subscribing to the 
Microsoft way of doing things.  We have other ways of telling the type 
of a file, why not let people name them whatever they like?  And I'm not 
asking for a ridiculous computer science reason...if we want to make the 
computer a true tool that anyone can use, we need to eliminate these 
ridiculous requirements that foil new users.

> There's also a nice little rant about Apple screwing up by forcing
> extensions into OS X:
>
> http://www.arstechnica.com/reviews/01q3/metadata/metadata-8.html

This is pretty much bull.  Apple is not forcing extensions on the user, 
but they are requiring that the application developer be able to save 
files with extensions, and read files with extensions.  MacOS X apps do 
not *need* extensions, but Apple had problems with win32 
interoperability in early builds which did not have extensive support 
for filename extensions.  If you are only using MacOS X, and no other 
system, you do not need file extensions.

--
Nick