View Single Post
  #69  
Old March 18th 18, 07:00 PM posted to alt.comp.os.windows-10
Paul[_32_]
external usenet poster
 
Posts: 11,873
Default Do you have an iOS device? How to get it to xfer screenshotsto Windows 10 easily without that iTunes abomination?

Uultred ragnusen wrote:
Paul wrote:



You can add file systems to Windows via the IFS mechanism.


I didn't realize I was "adding a file system" but if that's the same thing
as "mounting" a device, then I'm ok with that concept.


Linux has two flavors of file system mounting.

Well known, well tested ones, are placed in the kernel, in Ring0.

However, a lot of file systems, started as Ring3 prototypes,
which can be started as if they were applications. When run
as root, they allow file system mounting, for "foreign" file
systems.

Using such a mechanism, I could probably mount HFSPLUS for example.
At one time, the two Apple desktop standards, were reduced to crummy
little command line utilities, with limited capability. Now, they're
in much better shape.

So when you say the word "iFUSE" to me, I parse that it two parts.

i = some sorta iOS thing
FUSE = A ring3 mounting scheme

https://en.wikipedia.org/wiki/Filesystem_in_Userspace

The Windows IFS is a similar sort of mechanism. A person
with an iFUSE source code in hand, knows how to parse
some block device and make sense of it. You would transfer
that part of the info, into an IFS implementation on Windows.

What this notion of "stuffing" foreign file systems into
an OS entails, is "faking" the info for a stat() call.
For example, Linux has a value called inode number.
NTFS just happens to have an identifier called Filenum#.
Now, if you're writing a file system for Linux, you
stuff the "filenum#" value into the "inode" call,
and Linux goes away happy. It never needed the inode
value in the first place (because NTFS doesn't have
inode), but, if any application software comes calling
and runs stat() on a file, you must produce a value.
And perhaps, the value cannot "collide" with any existing
value. The NTFS filenum# is guaranteed to be unique (per file).

*******

Someone could "port" what was done for Linux, to Windows,
but... will they ? Will they charge money for it ?

There aren't really a lot of IFS implementations, which
suggests this is a moderately difficult topic. Mind you,
there might not have been a lot of usage cases either.
After you do one for EXTn, all that's left is doing one
for a FreeBSD disk :-) Ugh.

Paul
Ads