Re: [nfsv4] New version of sparse draft (draft-hildebrand-nfsv4-read-sparse-01.txt)

["J. Bruce Fields" <[e-mail]>]

On Fri, Oct 01, 2010 at 10:28:55AM -0700, Marc Eshel wrote:
>  I can claim that trying to read unallocated part of the file is and
> error :) but I see the point of complicating the error handling.  We
> can instead just add more information to the reply without giving
> the error return code.

The status is the only switch you can really take advantage of in the
read reply, so I don't think that's an option.  (The minorversion rules
prohibit completely replacing the xdr for an existing operation, which
is a good rule.)

> I think this a simple solution that should be agreed on as
> alternative if we don't have a more general solution that will cover
> the requirement for this one.
> This is a good example how we this group start with simple ideas and
> ending up with a conflicted and convoluted protocol that drags for
> years.

I think you'll find a separate READZ will be simpler if anything.

That doesn't force you to accept any extra bells and whistles if you
don't want to.

--b.

> We proposed it, I believe we got consensus. We can fine tune it to
> make every one happy. If someone has a better idea let them write a
> draft and if the group agrees that it is a solution we can switch to
> the new proposed draft.