> > So the only way I can > > think of that Firefox or Thunderbird might start > > faster after "several power > > on/off cycles"' is if those files they maintained > > happened to fall into a more > > efficient state, which might happen if they for > > example age them at startup > > time. > > Could you, if possible, elaborate on the "age" part?
I was thinking of a browser's web page cache. I would suppose it gets rid of old pages to stay within the amount of storage it is configured to use, and that it might do that at startup rather than freezing up once in awhile while running. > During the first couple of power on/off cycles, > Thunderbird was noticeably slow. It would take > probably five or more seconds to load. Now, it takes > only a couple of seconds. Very acceptable. No idea. Some mail readers are aggressive about looking for anything that even might be a mail folder; once they've got a reasonably current picture of things, they might settle down. Or there might be other reasons. I don't personally see most of that as a desktop issue (however much a desktop user might see it that way!) so much as an individual application issue, because quite likely it's the individual application that needs some work done to it to speed startup. > The biggest difference I have noticed is with > StarOffice 8. During the early cycles (i.e., after > it has been initially set up), it would take "a > while" (can't really quantity it, but it is > definitely much longer than five seconds) to load. > Now it takes less than two seconds. Needless to > say, I was initially very disappointed with SXCE 65, > but now I think I can say that I am quite > satisfied. I _think_ I might have seen this too, but I was never curious enough to investigate. The problem with anything like this is that to identify the problem and start working on a fix, it has to be possible to replicate the problem more or less at will. Loading a new system, or even creating a new account, and then collecting detailed information on the behavior of selected apps over the course of a few days, is fairly time consuming. Even the people who might know how to do that, let alone the ones who might know what to do _with_ that data, probably have lots of other things they already _know_ they need to be doing... > Also, something totally unexpected happened. The way > I have been using my SXCE box, every time after the > GNOME login, I always click on the JDS icon and > select "Firefox Web Browser". Then came yesterday, > when I found SXCE, sua sponta, added the familiar > Firefox icon on my GNOME desktop. I knew (though > never believed it) that Solaris kernel is supposed to > have some kind of self-healing power, but probably > not this kind of intelligence. > > Is this a feature, a bug, or what? No idea; I almost never use GNOME (still too slow :-) and I like CDE better anyway). It may have some limited smarts about finding certain apps and adding easy access to them, although I would have expected that to happen at first login, or be visible by second login at t he latest. Or, many desktop environments can "save" what programs are running, and restart them automatically the next time you log in. But they may only do that if told to, or if you log out a certain way, or otherwise engage in the proper magic. That seems to me a likelier explanation... Anyway, overall perception of desktop performance is apparently affected by a _lot_ of factors, not just involving the toolkit libraries and window manager, but all sorts of configuration options, individual application behavior, etc. It means there's not so much one magic solution as a bunch of things that need to be investigated, tweaked, or fixed. That takes a lot of time to get done, even assuming an organized effort is being made. (However, startup time of commonly used apps is an obvious enough target, I suppose.) This message posted from opensolaris.org
