From: Skybuck Flying on

Sometimes files/folder belong to different categories at the same, for
example: "graphics related", "music related", "programming related", "math

Currently windows requires me to put the file in folders, this creates a
"static" view of the situation. In reality I like placing all the
files/folder in one big folder so I can "easily" find everything.

However when the number of files reaches thousands or ten thousands it
becomes a bit harder to "sift" through it... I have no real way of
"filtering out" unwanted files, or "filtering/zoom into the files I wanna

So I recommend a new way of looking at files in the file system:

The user receives the ability to "tag" files. Which means the user can now
start adding "tags" to files.

So for example:
MyGameSource.cpp tags: "c++", "game development", "graphics", "game engine",

Then the next time I want to "sift" through my files... for example I want
to see all "graphics" related files...

So then I type the "tag" search term into the windows explorer like so:


The future windows/file explorer should filter out all other files which are
not graphics related.

Another example:

tag: "redcode" ;)

Search: "redcode"

All files related to redcode are shown...

This could be done for the current folder, there should also be an option
which says: "Search subfolders as well" and display those results as well !

Let's call this new idea: "Tag" based file browsing ? ;)

Skybuck :)

From: Art Overby on
> Let's call this new idea: "Tag" based file browsing ? ;)
> Bye,
> Skybuck :)

From: Skybuck Flying on

"Art Overby" <dont(a)> wrote in message
> >
>> Let's call this new idea: "Tag" based file browsing ? ;)
>> Bye,
>> Skybuck :)

I think best solution for windows would be to extend the ntfs file system
with tag support.

Each file/folder gets a special stream attach to it which can contain tags.

Then all that's needed is some extra interface work, some extra api work so
that the user can add, and search/filter on the tags.

So the main idea is to extend what is already there, so the user can keep
working they way he/she is used to but can try out the new tag/search/view
feature as well ;)


From: Skybuck Flying on
Anyway for people wanting to try it out in an implementation, maybe the
filedisk tool/technology/source code would be a good start...

What would be missing is api or maybe there is an api for extensions, and
the windows explorer would have to be re-done or extended...


From: Gavin Scott on
Aleksandar Kuktin <akuktin(a)> wrote:
> But I like the filesystem. :(

The UNIX file system paradigm is an abomination unto all mankind.

A file system which refuses to know anything about a file other than
the number of bytes it contains is like an OO programming language
where the only type you're ever allowed to use is "Object".

There was once a time when file systems tried to be helpful to
applications, remembering what sort of thing each file represented,
how it was arranged, etc.

But UNIX refuses to take responsibility for anything that might be
considered domain-specific. If not everyone needs it, then it's
inappropriate for the file system, the programming tools, or the
operating system to know anything about it. It's an application issue.

Very convenient if you're an operating system developer, not having
to bother in the least with any application of your system.

"We give you bags of bytes. From that you can derive anything."

It's not that you "can" do something with a tool like UNIX, it's
that very often you want to be sure the system "must" do something.

Very hard to do when the application has to take responsibility
for everything that a proper file system ought to, and the OS is
happy to let anyone or anything violate the constraints that the
application is trying to enforce.