[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-core-devel
Subject:    Re: Review Request: Fix sanitization of dbus path in KMainWindow
From:       "Matthew Woehlke" <mw_triad () users ! sourceforge ! net>
Date:       2009-08-11 15:25:27
Message-ID: 20090811152527.23576.59812 () localhost
[Download RAW message or body]



> On 2009-08-11 05:23:13, Chani wrote:
> > how will this affect applications that do have a - in their name? and \
> > applications that use the dbus interface of those apps?
> 
> thomasl wrote:
> As a dbus path attempt of "/some-app/MainWindow_1" is catched by an assertion in Qt \
> (for being invalid) there's never been a chance that you used it this way 
> If we'd however catch the appName in KAboutData to align the dbus path with e.g. \
> qApp->applicationName(), this would -likely- affect at least the KConfig rc path \
> and possibly the way it registers with KGlobalAction :-( 
> (if anyone wonders, i had of course a much cooler "." in my failing app ;-)
> 
> As the limitiation of the appName would be artificial, one could also choose to \
> change the dbus path system of KMainWindow to e.g. "/MainWindows/MainWindow_1", \
> "/MainWindows/MainWindow_2", etc. - but that would break everything here :-( 
> Tom Albers wrote:
> Concrete question: will:
> 
> QDBusInterface dbus( "org.kde.plasma-desktop", "/App" );
> 
> change into 
> 
> QDBusInterface dbus( "org.kde.plasmadesktop", "/App" );
> 
> ?
> 
> thomasl wrote:
> No. "plasma-desktop" is a valid /service/ and "/App" is a valid /path/
> If plasma-desktop had a mainwindow KMainWindow would (as by Matthews patch) \
> "change" it's /path/ to 
> /plasma_desktop/<objectName()>
> 
> as
> 
> /plasma-desktop/<objectName()>
> 
> would be an invalid dbus path (i didn't write the spec, ok ;-P)
> 
> fact is, we /need/ to fix up KMainWindow (or have developers figure out that they \
> have to change their appname when adding one) 
> As a consequence we'll live (in the plasma-desktop case) with one of the options:
> - "/app_name/MainWindow_1" != "/"+qApp->applicationName()+"/MainWindow_1"
> - qApp->applicationName(): plasma-desktop -> plasma_desktop
> - /app_name/MainWindow_1 -> /MainWindows/MainWindow_1
> 
> each introduces some kind of break, where the latter one have external impact and \
> the first (aside the devs need to remember) creates an inconsistency in the way to \
> access the KMainWindow dbus object -> :-(

This patch doesn't affect appName, just objectName. We may want to do something about \
appName also, but for the moment that is still an open question; please see \
http://permalink.gmane.org/gmane.comp.kde.devel.general/58753 .

Unless I misunderstand the spec, I don't think this can possibly break anything, \
since all it does is allow paths that previously would have been invalid (and would \
have thrown an assert leading to your application not working anyway).


- Matthew


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://reviewboard.kde.org/r/1261/#review1969
-----------------------------------------------------------


On 2009-08-10 22:27:09, Matthew Woehlke wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://reviewboard.kde.org/r/1261/
> -----------------------------------------------------------
> 
> (Updated 2009-08-10 22:27:09)
> 
> 
> Review request for kdelibs.
> 
> 
> Summary
> -------
> 
> As pointed out by Thomas Lübking ( \
> http://permalink.gmane.org/gmane.comp.kde.devel.general/58749 ), KMainWindow \
> attempts to sanitize what it will allow in the dbus path. However as written it \
> would allow the illegal characters "." and "-" to be passed through. 
> 
> Diffs
> -----
> 
> /trunk/KDE/kdelibs/kdeui/widgets/kmainwindow.cpp 1009802 
> 
> Diff: http://reviewboard.kde.org/r/1261/diff
> 
> 
> Testing
> -------
> 
> built kdelibs and ran a KDE application
> 
> 
> Thanks,
> 
> Matthew
> 
> 


[prev in list] [next in list] [prev in thread] [next in thread]