I am attaching the patch.

Btw I set ACK timeout to 5 seconds in this patch, but maybe we should consider 1second or even less (there is 1second resolution so it may be a problem).

In tftp_set_timeouts(...) function there is also a conditional statement that possibly is not needed:

  if(start) {
...
  }
  else {
    if(timeout_ms > 0)
      maxtime = (time_t)(timeout_ms + 500) / 1000;
    else
      maxtime = 3600;

As far as I understand, condition (timeout_ms > 0) is checked before "if (start)", because Curl_timeleft(...) function never returns 0.

> BTW, this change calls for a test case to verify that it does the right thing.
> Have you looked into how we can make one?

I don't really know how to write such test. Basically we need to simulate reception of unexpected blocks. Test for timeouts would be useful as well.

Marcin Adamski

-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html