Tidying up after Linux 16764

I understand this perfectly well. Just like .zip files begin with "PK, .gif files begin with "GIF87a" and so forth. But when I read one of the original posts it gave me the impression that this 'magic-number' stuff is some unique thing that is Linux specific. Somehow these 'magic numbers' were being inserted into the file so that the file could be identified by this number and this number alone.

I've mentioned (along with others) that this is how the Mac did it. There is a 'fork' in a Mac file and one branch has the raw file data (including file-format specific data like MZ, PK, GIF87a) and the other fork is typically a resource fork that contains meta-data as to what app created this file-type and how it should be opened.

It seemed that someone was implying that this sort of 'fork-like' feature was being used on Linux. I don't believe that the ext file system supports something like this so I was very curious where this data was being added without breaking several file formats. But from your post I now have the impression that there is no 'magic-number' that gets added anywhere to a file. 'Magic-number' is really a mechanism for identifying the file-type based on certain known file signatures-characteristics.

It's probably more of a compatibility issue than anything else but Msft could easily have implemented storing meta-data in a 'fork' exactly how it's done on a Mac. From the start the NTFS file system has supported "multiple data streams" which is basically a file-fork. On NTFS the name of the file ("Foo") is the primary data stream. "Foo::type" could be the 'type' of file and other streams like "Foo::author" and "Foo::MyPetsName" could also exist. I suspect that wide usage of FAT16 and FAT32 (or perhaps plain stupidity) prevented them from moving in this direction to identify file-types.