Re: What are the accepted values for "Dialog Toolkit" in the latest betas?
[email protected] Sat, 08 Jan 2011 10:54:49 +0000
| Newsgroups | gmane.comp.web.opera.linux |
|---|---|
| Message-ID | <[email protected]> |
Quoting Kenneth Crudup <[email protected]>: > > On Sat, 8 Jan 2011, [email protected] wrote: > >> Opera will attempt to load KDE libraries on startup in mode 3 (and in mode 0 >> if it detects an active KDE window manager). I do not remember exactly which >> one it needs right now, so I would recommend to install a minimal >> KDE setup as >> that will pull in all required libraries. That may be a bit overkill for you >> though as I assume you do not use or need KDE for anything else. > > Thanks for your reply. > > I'll experiment with this, but *please please* try to see if you can > find out what it's looking for- the memory overhead by having KDE's libs > around is *far* less than the effect of the Gnome/GTK file selector after > saving a few tens of files; Opera 11 is now stable enough that it doesn't > crash, yet I have to shut it down every day or so 'cause that file-selector > leaks so badly my machine ends up thrashing 'cause the resident size is > so large. > > ... as an alternative, would it be possible to just not have the fallback > X11 file-selector *not* stat(2) every file it sees in a directory? How many files and/or sub directories do you typically have in "problematic" directories? I assume the number is high since you mention it. Is it a regular local native file system? It might be that it is not stat'ing alone but also filtering and populating the list that takes time. One solution might be to cache the contents of visited directories and add some sort of monitoring to update the content when needed in the background. I assume you know you can save linked content to a predefined download folder without using any file dialog. -- Opera-Linux: https://list.opera.com/mailman/listinfo/opera-linux More lists: https://list.opera.com/mailman/listinfo/ Unsubscribe: mailto:[email protected]?subject=unsubscribe