Re: abstract file (support multi-part)

[Clayton Weaver <cgweav _at_ eskimo.com>]

(whoops; oh well)

>   > So, in sum, "everything is a directory"?
 
> While I agree with the majority of what you said, this leads to confusion;
> multi-part files may have similar semantics to directories, but calling
> them such complicates the issue.
 
> I know I keep harping on about this, but it *is* important -- it's
> difficult enough getting new users to use the correct terminology for
> things without the people who develop the stuff getting it wrong too :)
 
Sure. Working from this model, what you have is an abstract hierarchical
namespace, a tree with arbitrary fanout, and "directories" are just one
user-space representation and semantics for a part of the tree. How
the kernel handles them is an implementation detail. Userspace doesn't
know what "dentries" are, so why would it need to know the details of
a more abstract hierarchical storage object implementation?

Note: remember the dentries development process. Reimplementing VFS to
work from a multi-part storage object as the default model and keeping
everything that works now working is not going to be some trivial
weekend hack, without even considering the performance implications.

(ref: "Everything is a file" in explanations of the traditional unix
api model; "everything is a directory" refers to an implementation
model rather than an api model.)

Regards,

Clayton Weaver
<mailto:cgweav _at_ eskimo.com>
(Seattle)

"Everybody's ignorant, just in different subjects."  Will Rogers



-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo _at_ vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/