In article <jfjd2494fskclils1vu5dtqu3ocpvj73cf@4ax.com>,
Brian A <no_spam_bca1000@hotmail.com> writes:[color=blue]
> On a separate topic, not wishing to hijack the thread...
> I remotely control one of the SPA-3102 so that I can sort out any
> problems from home. I have found that it is impossible to get the web
> 'Voice' control interface to come up correctly on a Windows XP Pro
> platform using Firefox, Opera or MSIE 7.
> The problem I get is that all the tab contents displayed under each
> other, in one long list, and then the egg timer just hangs as if
> waiting for more data. I have tried emptying the RAM and disk cache to
> no avail. I am aware that this web interface uses javascript so there
> may be some problem there though I don't think that is the problem.
> In Ubuntu Linux it will display OK most of the time but sometimes the
> page needs to be loaded again for a correct display. So, I am tied to
> booting up in Ubuntu if I want to access the SPA-3102 remotely.
> Any ideas on this one?[/color]

It's a bug in the TCP stack in the Sipura's. It was
reported to Sipura years ago, but seemingly not fixed.
If the outstanding unacked data gets to 32kbyte, the Sipura
messes up it's calculations of where it's got to and starts
retransmitting already acked TCP data. (This is likely a
classic coding mistake, treating the window value as a signed
16 bit value rather than unsigned 16 bit or a 32 bit value.)
The remote end will give up and reset the TCP connection,
leaving you with a partly loaded web page, which causes the
tabs not to work (or in some browsers, no display at all).

This is most likely to happen when accessing the web pages
across a WAN (where there's a transit delay), and when
accessing the larger web pages (e.g. admin, advanced).

To workaround this, set the maximum TCP receive window on
your browser system to 16k, so that the Sipura can never
get to 32kbytes of unacked data outstanding. How you do
this will depend on your OS. If (as is normally the case)
the setting applies to all TCP connections, it may have an
adverse effect on TCP performance in some circumstancies.

--
Andrew Gabriel
[email address is not usable -- followup in the newsgroup]