I spoke to some users and they said that the freezings are related to the TCP/IP protocol. It should be true, because they happenned only with browsers.
A person suggested me to install a proxy. I found the "dproxy-nexgen" which is very lite and is a port of Paul Smedley.
Does anybody know how to install it? The documentation doesn't explain much.
Thanks in advance.
Hi...
The WPS doesn't freeze due to the TCP/IP stack (as far as I have seen)... I run servers that crank out gigs a day and the WPS is always responsive (till I launch Firefox - or OO -- killing/shutting them down resolves that issue). The way the rendering code is set up for Firefox is still an issue with OS/2, and bug tickets have been out for years, but no one has managed to track down the cause of the problem the last time I checked (which is why you can watch Firefox render, render, re-render pages in OS/2, while an XP machine will be blazingly faster at it).
Of course, the servers are DominoGo 4.6.2.8, MySQL, FTPDServer, and OS/2's DNS Server... which were written for OS/2... and only DominoGo has a GUI interface - and that interface seems to run in a separate thread/process.
I have a feeling that the separation between the GUI components, and the networking components in certain Windows ports aren't as well separated... but if that's the case, it's not the TCPIP stack - but the "poor" (ie: different) design of the Windows ports. OS/2 apps (if you check the programming tutorials that come with stuff like VAC++) are supposed to process GUI messages separate from other program tasks to prevent the GUI from experiencing delays due to unprocessed messages (or some such concept)... I dont think that design method applies in the same way to Windows apps... hence issues with Windows ports... hence something with a TCPIP component could theoretically "hang" the GUI due to the app not processing it's messages in a timely fashion while waiting on some sort of TCPIP traffic.
One other thing I noticed is that with VIO apps, the GUI (well, everything) will slow down if it is displaying it's activities to the VIO window... but that is most noticable (in my experience) on the machines I have built using nVidia video cards and SNAP (as the nVidia cards dont seem to be supported very well). I've got no such issue with the ATI's I have... I have ancient ATI cards that run circles around brand new nVidia cards
(on OS/2), and doing anything VIO display intensive (a directory listing like DIR /S) in a VIO window will peg the CPU (until the window is minimized or rolled into it's titlebar or covered).
-Rob