Comment Re:file magic - use the content to determine type (Score 1) 564
We're talking at cross-purposes.
Possible, but I see you are factually wrong.
My view is that we shouldn't be identifying files manually AT ALL.
No one in the whole thread is advising identifying files manually. Different people are proposing file names, extensions, content etc. all non-manually / automatically / programmatically / transparently to the end user.
They should be part of the meta-data, as already is whenever you download a file. Just because it ends in
And I never said putting this information in file name is a good strategy, so why you litter your post with such irrelevant content is beyond me. This is why I read but am not addressing more than half of this post of yours too.
In fact, you can break stuff easily that way if you don't populate your webserver with proper mimetypes.
I got it. You are among the proponents of evil bit. I agree information security is trivial once we get all the evil people to set this evil bit. Real life is not so simple - web servers areone of the least trustworthy elements in a typical user's computing life.
But from that point on, we don't NEED to ever identify a file again
In RFC 3514 world, yes. In real life, this "metadata" will need to be edited, or distrusted. So we do need to identify a file. In the face of this imperfect world, there are certain difficulties. Whether it is a perl script or a jpeg image can best be figured out from looking at file content.
NOT from file name.
NOT from metadata set by untrustworthy people.