file types/data types/mime types

I am puzzled about something with osX.
Why does it use the file extension for determining file type instead of
looking into the file for the type?

My Amiga used datatypes 15 years ago to determine what kind of file it
was. If it was jpeg but named .gif, no problem. The jpeg program loaded it.

For instance, I had a file on my desktop, an html file of instructions.
It had no file extension so appeared on the desktop as a blank white
rectangle. Re-named it 'html' so firefox showed up on the icon.

I just renamed it 'jpg' and GraphicConverter tried to open it. It
reported "The file 'foobar.jpg' is broken, the format is unknown or it
isn't a graphics file.
So, Graphic Converter can't open it."
Then it gives a file dump, in hex and ascii which clearly shows it to be
an html file.

Why aren't data types or mime types implemented in the user interface?
Are datatypes/mimetypes in the BSD itself? Cannot find it.

 

I don't know the 'old type/creator' system as I hardly ever used a Mac
prior to osx.

I'm talking about the file type data which IS inside the file. Like this
jpeg, says "ﾘ…�

 

I think you're referring to the "magic number" scheme which attempts to
identify the type of a file using some text embedded in the file's
contents. The scheme works in Mac OS X as well as it ever did (see the
man pages for "file" and "magic") but fails when the file doesn't adhere
to the requirements or the name refers to a directory or bundle.

 

Most files do not have file type data inside the file.

 

Unfortunately, there is as yet no standard for including such
information in all types of computer files. Until there is, identifying
file types by their content will be hit or miss. Perhaps the move toward
standardized file formats will lead the computer industry to come up
with a standard for including basic metadata within files themselves so
that they can be moved among different file systems without having to
worry about filename extensions and such.

 

It turns out that it is difficult to achieve this, because
computer programs often do not process whole files, but some substrings.
Then the type information is lost anyway. One example, discussed much, is
the use of the BOM sometimes put in the beginning of files to tell whether
it is UTF-16 big or little endian. It does not work well in a UNIX
environment, just causing problems in an encoding such as UTF-8.

So it is unlikely that metadata will move into the files for general
processing, but only to be used in special formats, as is the case now.