[U-Boot] [PATCH 0/4] Network defrag

[Robin Getz]

On Wed 12 Aug 2009 17:30, Wolfgang Denk pondered:
> Dear Ben Warren,
> 
> In message <4A832BCE.9060100 at gmail.com> you wrote:
> >
> > Sure, if you don't mind re-compiling.  I think it might be an 
> > opt-outable message via puts_quiet()
> 
> It seems we start having a mess here, with features bound to other
> features that have not even been agreeds about yet.
> 
> I have to admit that I am no friend of this puts_quiet() thingy. 
> How much time do we really save on a normal system?

A small fraction.

On a high speed network, with a low speed UART, and block sizes of 512 
(default with BSD tftp).

With printing hashes...
uart    Download   
57600   6657 (ms)    2,734,393 (bytes/sec)

Without
57600   6418 (ms)    2,836,219 (bytes/sec)

As the network gets faster, and the UART gets slower - it will make a bigger 
difference.

> Is this worth the inconsistent behaviour

Most likely not. I'm not going to be offended if you NAK it.

The better thing to do (IMHO) - would be to print out the proper number of 
hashes, depending on the size of the file (and implement RFC 2349 at the same 
time) - not the number of packets (which is what happens today)...

> and the (IMHO much uglier) code? 

The code isn't uglier - one function is replaced with a different one.

The name of the function is kind of crappy - but that is what I came up with 
late one evening...