Il lun, 2003-09-29 alle 12:53, Ettore Perazzoli ha scritto: > On Wed, 2003-09-24 at 14:09, Rodney Dawes wrote: > > The following patch changes the shell to register with the display > > information, so that multiple logins to the same machine with different > > displays, will work properly. Everything seems to work. I tried > > switching between folders, opening the settings dialog and going through > > the different pages, and composing mail. They all seem to work fine. > > This is not the right way to do it; you still want one shell to be > running at all times, because that's an assumption that both the shell > and the components make. Why is it making that assumption? It seems like a rather poor assumption to be making anymore, since all the components are shlibs now, and must be loaded by the process anyway. > So you don't want a shell process per display, but rather, one shell > process that has windows on multiple displays. I'm not so sure that this matters as much with the new shared library components. > I outlined how it should be done here: > > http://www.mail-archive.com/evolution-hackers lists ximian com/msg00678.html This thread also mentions locking, for the case of multiple machines accessing the same shared home directory. And given the probability that you would be writing to the exact same file from two different machines, at the exact same time, with Evolution anyway, I would say that locking is less of an issue, than updating all the currently open views. The same case also makes sense for multiple displays on the same machine. And, in this case, there is still only one instance of the wombat running, since it does not require a display, all instances of Evolution by the same user, can access the same wombat. And, Camel already does locking on the mail folders. So, all in all, this seems like a sufficient fix/workaround for 1.4, and will make many people much happier. I won't disagree that we should be doing more than that, also, but this seems like a more timely, and useful fix for 1.4. Perhaps we can do the right thing for 2.0. It seems like more and more people are asking for it, especially in the cases where thin clients seem to be coming more prevalent. -- dobey
Attachment:
signature.asc
Description: Questa parte del messaggio =?ISO-8859-1?Q?=E8?= firmata