Differential D24945 Diff 68751 ksmserver/org.kde.KWin.Session.xml

Changeset View

Standalone View

View Options

ksmserver/org.kde.KWin.Session.xml

  • This file was added.
1​<!DOCTYPE node PUBLIC "-//freedesktop//DTD D-BUS Object Introspection 1.0//EN"
2​"http://www.freedesktop.org/standards/dbus/1.0/introspect.dtd">
3​<node>
4​ <interface name="org.kde.KWin.Session">
5​ <method name="setState">
6​ <!-- set state. One of: normal, saving, quitting -->
7​ <arg name="state" type="s" direction="in"/>
apolUnsubmitted
Not Done

Here it recommends using uint for such things:
https://dbus.freedesktop.org/doc/dbus-api-design.html#interface-files

apol: Here it recommends using uint for such things: https://dbus.freedesktop.org/doc/dbus-api-design.
davidedmundsonAuthorUnsubmitted
Not Done

It also advocates for use of {sv} in your spec which is inherently string based.

I'm happy to use ints, as long as we then don't get into that trap of saying; "now we have an enum to sync in two places, we must create a header file and a dependency "

davidedmundson: It also advocates for use of {sv} in your spec which is inherently string based. I'm happy to…
davidedmundsonAuthorUnsubmitted
Not Done

There was a third option I considered. Have N methods

startSave
startQuit
saveCancelled

and then make the state change an implicit part of the invokable action we need to do on save.

I went this way as I felt it was nice to have this acting more like a property independently, but what do you think.

davidedmundson: There was a third option I considered. Have N methods startSave startQuit saveCancelled and…
apolUnsubmitted
Not Done

I guess it doesn't matter ultimately. I won't advocate for a header with an enum, don't worry.
Logically, having an int is simpler than a string and we will have to document both and have an implicit tie in both implmentations.

Having 3 methods seems quite elegant as in it's a more semantic API.

As you put it (t, the 3rd seems simpler to grasp, not that everyone should be looking at it.

apol: I guess it doesn't matter ultimately. I won't advocate for a header with an enum, don't worry.
8​ </method>
9​ </interface>
10​</node>