Desktop Entry Specification - ExecuteAs proposition

[PCMan]

On Sun, Mar 7, 2010 at 10:03 PM, Dario Freddi <drf54321 at gmail.com> wrote:
> On Sunday 07 March 2010 14:54:16 you wrote:
>> On Sun, Mar 7, 2010 at 8:51 PM, Dario Freddi <drf54321 at gmail.com> wrote:
>> > I am not a big fan of this patch because (especially given what I've
>> > written above) I don't see a use case.
>>
>> You personally don't use it doesn't means that others don't need it or
>> there is no use case.
>> It's widely used in SuSE and it's required by some Debian/Ubuntu packages,
>> too. It's especially useful when you have a simple program and the part
>> dealing with dbus + policykit support is even bigger than the program
>> itself. That's the use case.
>
> By the way, this is relevant for GUI programs, proof that you didn't read my
> mail at all, and not for small applications. I doubt a GUI program doing some
> stuff as root can be _that_ small, but I might be wrong.
For some smaller systems focusing on simplicity, it's possible to have
GUI programs of this size. Of course if you're talking about something
whose complexity is at the level of packagekit, policykit is a better
choice. How about some simple GUI program written in zenity or pygtk?
I mentioned this in previous mails and hope you have read that.
If you just need the users to enter something, and save those things
to config files, this is a good use case. For example, a small
configurator for pppoe or something can be this size.
For some small systems where Gnome or KDE is not installed, it's
possible to have some simple and quick config tools of this size
written in scripts. For developing large GUI apps, of course people
will consider policykit seriously and not trying to run the whole GUI
using root. That's why I mentioned small programs since most of the
time that's the realworld use case.
>> As I said previously, it's a matter of ``choice''. Developers can use
>> policykit if they want, but don't prevent others from using sudo
>> simply because you don't use it. Adding this key doesn't prevent
>> developers from using policykit. It only adds usefulness for those who
>> need it.
>>
>> The "No, you don't need that" strategy is not very good for a healthy,
>> open, and free community.
>> Issues are raised again and agina simply because people need it. The
>> best way to deal with this kind issue, IMHO, is not "No, you are
>> wrong. You don't need that.".
>>
>> Maybe it's not needed for Gnome and KDE, but our users in other
>> desktop environments need it. Otherwise we shouldn't get this kind of
>> feature requests and user feedback.
>
> Please calm down. I don't represent neither GNOME or KDE. I said that I, as in
> me and nobody else, don't see an use case, and it's my opinion. If you were
> looking for people saying "it's ok throw it in", and not even listening to
> what I said (you quoted only the last part of the mail, which is the less
> relevant one), then why are you discussing this patch in the first place?
>
> Here you're just stating what you think is right without even listening to
> others.
I read all of the mails before replying, including yours.
> Respecting other people's opinions is very good for a healthy, open, and free
> community.
I just provided some use cases. Most of the time this kind of programs
tend to be small and simple so I mention them specifically, not
because I didn't read your mail. For other complex and sophisticated
software, I totally agree the use of things like policykit and if I'm
writing this kind of software, I'll use policykit, too. But the
problem is there still exist some use cases for which policykit is not
the best tool and that's why we're here to discuss this patch.

Please accept my apology if you felt offended. Solutions existing in
freedesktop.org are good, but the problem is, there's no
one-size-fits-all solution.