RE: [lwip-devel] byte order, packing, optimizations

[Bill Auerbach]

>Personally, I'd like to see assertions or checks of some type if we
>remove structure packing from any structs that are exposed in headers

Assertions would be good if even just short term.

I think (hope) we're in agreement that struct padding of 2 should be the
default (over 1) since there is benefit in doing so and there is no need to
ever pack by 1.  A comment might be nice stating padding of 2 or 4 is only
necessary if the processor has 16 or 32-bit alignment requirements.

>The patch you attached does far more than that! And it's not clear to me
>if/why we need all that. Also, it interferes with / task #10173
><http://savannah.nongnu.org/task/?10173>/ (Add IPCOPY and HWACOPY):
>http://savannah.nongnu.org/task/?10173

I wondered if his patch in fact covers 10173.  And where does it leave us
with IPv6?

Speaking of that, it appears that uIP is alive and well and has been updated
as a part of contiki and is IPv6 compliant!
http://www.sics.se/~adam/contiki/docs-uipv6/

Sorry for my tone earlier, which was a bit obnoxious - I'm only hoping to
promote the idea that the more resources that are put out there to make lwIP
easier to port and/or use, the more that lwIP will actually be used.  It
follows that the more it's used, the better it becomes.

Bill