In message <[email protected]> Dave Higton <[email protected]> wrote:
> In message <[email protected]> > David Pitt <[email protected]> wrote: > > >Geoffrey Baxendale, on 4 Feb, wrote: > > > > > >> Is it just me, or is Netsurf's sockets system awry. After a modest > >> browsing session Internet access freezes up due, I think, > >> to sockets being used up. Qitting Netsurf restores order. It seems to > >> have started about a couple of weeks ago. Not sure of the exact time. > >> > >> Risc Os 6.2 NS R1681. (JSoff) > > > >The same thing happens here with #1694 on a Raspberry Pi, OS5.21 (31Jan14). > > > >I have resurrected a Socket Counter application, !SockCtr, which dates back > >to the time when Oregano2 managed to do a similar trick. It is a frontend to > >David Ruck's SocketCount. > > > >http://pittdj.co.uk/software/index.htm > > Thank you for that useful app, David. It confirms Netsurf's problem > nicely. The number of sockets starts off for me at 90, and gradually > collapses. When it reaches 0, NS can't resolve the link. The number > rarely increases at all. > > But the clincher is: quit NS, and the number instantly goes back up > to 90. > > I'm confirming it with build 1694 and 1615. There's a big gap in > the versions I've kept; the latest before 1615 is 1356, which behaves > very differently. Curiously, the sockets are recovered not when a > page closes, but when a new one is opened. > > Dave > As far as I can tell, 1614 is OK and 1635 not. Don't have any builds between those here. When I have more time I will try and narrow it down further. (All Non JS) TTFN -- Geoff. Using Acorn StrongARM Kinetic RiscPC. Bear Philosophy: "The majority are probably wrong"
