why NTFS reports incorrect file sizes

[D. Hugh Redelmeier]

| From: James Knott <james.knott-bJEeYj9oJeDQT0dZR+AlfA at public.gmane.org>

| Extended attributes were used in OS/2 and could support up to 64KB of meta
| data.  Windows NT could run character mode OS/2 1.x apps, so that's probably
| why extended attributes were supported in NTFS.  Microsoft also developed the
| HPFS file system, used in OS/2, back when they were partners with IBM in
| developing OS/2.

What were they actually used for?

As far as I know, there is no strong culture / convention for using
this capability in Linux.  As far as I'm concerned, all it
accomplishes is to break the "a file is a bucket of bytes" model of
UNIX files.  So tar won't work as a backup, cp won't work to copy a
file, etc -- many utilities are broken or need(ed) revision.

Extended attributes surely don't matter in Linux since I've been able
to ignore them up to now.

The beauty of UNIX compared with its precursors was simplicity.  I
moved to UNIX from IBM OS/360.  Files there had all kinds of
attributes that optimized how I/O was performed but actually just made
file I/O complicated.  Files had "record formats" (how file blocks
were to be broken into records), block sizes, record sizes, printer
control types, indices, and more.  UNIX had "just a bucket of bytes"
(plus, I admit, modest, fixed, simple metadata).

In MacOS (pre-OSX) the "resource fork" of each file was important and
there were strong conventions on how it was used.  So much so that Resedit
(the resource fork editor) was a very powerful tool for customization
without needing access to source code or being a programmer.

I don't know how OSX resolved its twin heritages.
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists