[lwip-devel] [task #7891] FTP Client for lwip ready for test

[Iordan Neshev]

Follow-up Comment #7, task #7891 (project lwip):

>why did you write your own socket layer? 

1. Why did we not use the existing socket API?
Because it needs OS. The current project takes
many features from older products and it would
take much time to rewrite them for an OS. I guess
we could make all our previous soft work as a 
single thread but did not want to risk when the
time-to-market is short. Also, every OS adds some
overhead. At the present time we are struggling
for RAM. The lwip's PPP needs much of it.

2. Because (almost) all the freeware in C around the
world is written for sockets (just like the eCos
client). If we want to benefit from freeware we need socket
API. For the current project we don't need other
network application than ftpc but for future project
we should do. Why rewriting every freeware application
for the raw API when we can implement socket API once
and use freeware with less efort and modifications?

3. Because I've never done network programming before.
This was the perfect opportunity to learn and
work at the same time.

>Isn't the lwIP raw API more efficient?
I'm sure that a client, based on the raw api
will be much faster, but since we use
GPRS connection this is impossible. 

Btw we want to contribte also with the PPP stack
modified for use without OS. Should I start a new
Task or add it to some older one?

    _______________________________________________________

Reply to this item at:

  <http://savannah.nongnu.org/task/?7891>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.nongnu.org/