2012/10/20 Doug Ewell <doug_at_ewellic.org>:
> When a Major Software Company, which sells the Well-Known Operating System
> that I and a few other people use and develop for, decides to add
> character-encoding metadata to the file system of that OS
This already exists as a builtin functionality of the filesystem
(except that it is not used for this kind of metadata).
These are called "ADS" (Alternate Data Streams), and it mostly used
today (and extremely frequently) for tracking the origin of the
content of the file from another domain (mostly Internet). The storage
filesystem in question offers a nice way to store them very
efficiently, in a very compact way (just like this filesystem also
stores very efficiently small file contents, instead of garbaging full
data clusters), and with fast access methods and indexing.
On MacOS, these were called "resource forks" (distinguished from "data
forks" for file contents).
The concept of ADS also exist on other filesystems (including in
Windows itself for its "registry", which is another kind of
filesystem, or historically as well in VMS whose filesystem, along
with the one for MacOS, plus UFS on Unix for "links", largely inpired
NTFS). They are not new, they already exist, but they are just
underused (except in HTTP where it is builtin and widespread) :
We are too much used to the old hierarchical model for filesystems,
even if multiple streams can perfectly exist at every level, and have
their own separate structures (not necessarily hierarchical).
Received on Sat Oct 20 2012 - 18:08:41 CDT
This archive was generated by hypermail 2.2.0 : Sat Oct 20 2012 - 18:08:42 CDT