AW: [Oberon] Re. WinAos networking: patch available

[Stauber Sven Philipp]

Hi,
 
>2. What's wrong with USB - also for Win & linux ?
>Why is there still much debugging and discussion, on various
>forums although it's several years old already ?
That's because of the "U" in USB. The philosophy behind USB is to provide a single solution which is able to handle almost any kind of devices (low-/high-cost, low-/high-speed, simple/complex,...) on a wide variety of platforms (for example, USB provides quite sophisticated power management facilities to be also an appropriate solution for the use with battery-driven hosts). The same time, it should be easy-to-use "plug'n'play". This makes the system more complex than device-specific solutions are. The core system (UHCI, OHCI & EHCI host controller, USB bus driver) is specified in almost 1000 pages (and there are MANY more specifications when also considering the generic class drivers (see www.usb.org)).
I haven't seen another bus system which is used by a such great variety/quantity of devices.

Regards,
Sven

________________________________

Von: oberon-bounces at lists.inf.ethz.ch im Auftrag von easlab at absamail.co.za
Gesendet: Sa 19.11.2005 15:22
An: oberon at lists.inf.ethz.ch; easlab at absamail.co.za
Betreff: [Oberon] Re. WinAos networking: patch available




Felix Friedrich wrote:
> in the Windows Oberon versions the Winsock built-in asynchroneous
> networking was used because the single threaded system would have
> been blocked by waiting for connections or new data to arrive.
> ... snip ..
>
> Fortunately, Edgar Schwarz gave me some useful hints concerning
> ioctlsocket  (etc.) and we could therefore get rid of the circumstantial
> way of using buffers for sending and receiving while Windows has its own
> network buffers. I think that networking under WinAos is faster and more
> reliable now.

I want to raise 2 issues potentially related to WinAos networking.

1. Probably WinAos networking evolved from NO networking ?
  Theoretically some subtle design flaws could be inherited.
  It seems near certain now that my repeated reports of
  apparent design flaws in NO > ppp is simplistic/wrong.
  Now I'm regularly able to get and clear the problem.
  Which seems to be in the serial drive to the modem:
  ppp is 'integrated with' the serial-driver [V24 module].
  As previously noted a similar problem is that the V24.Panel 's
  "Open" is ineffective under certain conditions. [things always
   work if you just boot and test "it", but real life means that
   unanticipated environmental conditions exist when at unpredictable
   times you want  to do "it"].  When the problem occurs [eg. because
   my ISP doesn't  log-me-in, apparently due to too much load;
   {I still can't see why they don't feedback 'engaged' and save
    subscribers a call-cost} ], I fix it by:
       CRGV24Debug.CloseCOM2 ==> V24.Stop(1)
   This usually also causes a trap - which I'll investigate one day.
   Has this problem been inherited by Aos ?

2. What's wrong with USB - also for Win & linux ?
  Why is there still much debugging and discussion, on various
  forums although it's several years old already ?
  When will it be considered matured & stable ?

 == Chris Glur.


--
Oberon at lists.inf.ethz.ch mailing list for ETH Oberon and related systems
https://lists.inf.ethz.ch/mailman/listinfo/oberon