[En-Nut-Discussion] RFC: Something for the "next" TODO list?

[Michael Jones]

Hi,

I would like to purpose that we consider removing the "early buffer releases
on network errors" within the network code. As Harald explained to me once
that these constructs where intended to speed-up memory heap recover as
earlier boards were challenged by memory limitations. As we are moving to a
time where we have got a bit more elbow freedom memory wise, I suggest
taking on the task for the next release. I have done a bit of this myself
already and have only kept a low profile on this due to the old "breaking
3rd party software implications". Yet I believe that it should not be that
bad as this would not affect application code but "only" some drivers or
protocol implementations. I have found that these changes greatly enhance
readability of the protocol code and also limit the problems with forgotten
or duplicate buffer freeing.

Comments welcome!

Cu,
Michael