home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.amiga.programmer
- Path: peer-news.britain.eu.net!liv!lucs!news
- From: fish@csc.liv.ac.uk (S.E. Morris)
- Subject: Re: Protection bits
- Sender: news@csc.liv.ac.uk (News Eater)
- Message-ID: <DKyoqJ.Lnu@csc.liv.ac.uk>
- Date: Wed, 10 Jan 1996 10:49:30 GMT
- References: <john.hendrikx.44q1@grafix.xs4all.nl>
- Nntp-Posting-Host: fish@bobr.csc.liv.ac.uk
- Organization: Computer Science, University of Liverpool, UK
- X-Newsreader: knews 0.9.3
-
- In article <john.hendrikx.44q1@grafix.xs4all.nl>,
- john.hendrikx@grafix.xs4all.nl (John Hendrikx) writes:
- > In a message of 06 Jan 96 Michael Van Elst wrote to All:
- >
- > >> I'd like to see a new set of flags, or somekind of ID stored with each
- > >> file which tells you the filetype. In other words taking the 'e' and 's'
- > >> bit concepts one step further.
- >
- > MVE> With flags being pretty independent of the data in the file it is hard
- > MVE> to make them a file type ID.
- >
- > Maybe the filesystem could be extended with somekind of hook which sets the
- > file-type correctly by looking at the first 512-bytes of the file when it is
- > first written/modified. It would be an automatic FileID-ing system which could
- > be extended with new FileID's which the file system could recognise and store
- > as a FileID long in the fileinfoblock.
- >
- > When I come to think of it this doesn't seem all that hard to do and it would
- > probably not even slow down the filesystem. I mean, taking a peek at the first
- > 512 bytes as it is written shouldn't be that hard. The only trouble is that
- > 512 bytes might not be sufficient to make a FileID...
- >
- > This would make doing a directory with FileID's and with only IFF files in them
- > really fast, like on the MAC's (even though they use an even more extensive,
- > but unfortunately not always reliable system (you sometimes need to set the
- > FileType correctly yourself...)).
- >
-
- Why not store this infomation in the file's .info file?
-
- This kind of auto-recognition of files isn't a feature you would
- expect a command line shell to have --- it's a bit too high
- level for that IMHO.
-
- But the GUI is intended as a 'high-level' interface which goes out
- of it's way to be friendly and helpful (supposedly!). So storing
- a field in the .info with the file type would be ideal. The Workbench
- *has* to load the .info files anyway --- so the speed descrease would
- be practically non-existant!
-
- Just a thought.... :->
-
- -FISH- ><>
-
-