[pnfs] [PATCH] Removed sessions hard-coded 32678 byte max sizes

[J. Bruce Fields]

On Fri, Mar 16, 2007 at 02:02:46PM -0700, Iyer, Rahul wrote:
> The way I see it, the only thing left in the sessions code is:
> 1. Server side slot table
> 2. Server/client side callback code
>  (Anything else?)
> Since we're already at 14K lines, these two features might add another
> 2-3K maybe. If we're gonna wrestle 14K, then 16K might not be a lot more
> effort. 

Is it possible that we're already over the "may as well start over from
scratch" level of complexity?

Looking back at the 2.5-based NFSv4 patches, which I think are the ones
that were submitted (or at least close to it):

	http://www.citi.umich.edu/projects/nfsv4/patches-2.5/

The total diffstat was about 8000 lines changed over 44 files.  In
practice I think it was close to a complete rewrite of the initial
2.4-based prototype--this is a little before my time, so Andy would
remember better.
  
It was split into 38 patches, and it looks like they kept it to that few
only because though they managed to cram most of the changes into a few
really big patches.  It was a ton of work just getting to that point.

> That being said, I'll start working on the callback code as soon as I
> get the slot table patches split so Andy can merge them. After that I
> can start the migration to the latest kernel. I should be done (tested
> and stable) by the end of April, I think.

The business of splitting it up into easy-to-review-chunks is going to
be a lot harder than just a port to the latest kernel.

Well, I don't know what the right thing is to do.  I'm just worrying out
loud....  But I think we'd all worry less the earlier we could start
merging patches into upstream.

--b.