Review Request 125655: Xembed SNI Proxy
Martin Gräßlin
mgraesslin at kde.org
Fri Oct 16 12:34:08 UTC 2015
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://git.reviewboard.kde.org/r/125655/#review86912
-----------------------------------------------------------
Overall looks good! General comments:
* please use KSelectionOwner
* please disable cast to/from ascii
* please add unit test (yes I know, difficult, but should be possible)
xembedsniproxy/CMakeLists.txt (line 50)
<https://git.reviewboard.kde.org/r/125655/#comment59768>
are you sure you want -Wall. The last project I tried to build with that flag was unbuildable.
I'm mostly concerned that this causes problems when compilers start to add more warnings and perfectly no-warning code starts to fail to build for other devs.
xembedsniproxy/CMakeLists.txt (line 67)
<https://git.reviewboard.kde.org/r/125655/#comment59767>
variable called kwin_XCB_LIBS?
xembedsniproxy/fdoselectionmanager.h (line 31)
<https://git.reviewboard.kde.org/r/125655/#comment59769>
you are aware that we have a class in KWindowSystem doing that? see http://api.kde.org/frameworks-api/frameworks5-apidocs/kwindowsystem/html/classKSelectionOwner.html
xembedsniproxy/fdoselectionmanager.h (line 44)
<https://git.reviewboard.kde.org/r/125655/#comment59772>
suggestion: Don't use WId. Use either xcb_window_t (as those two below) or uint32_t or quint32. WId is just something weird for multi-platform.
xembedsniproxy/fdoselectionmanager.h (lines 50 - 51)
<https://git.reviewboard.kde.org/r/125655/#comment59773>
same here, you want xcb_window_t
xembedsniproxy/fdoselectionmanager.cpp (lines 51 - 55)
<https://git.reviewboard.kde.org/r/125655/#comment59770>
the reply->first_event should also only be read if the reply->present is true.
In addition: what's the strategy for damage extension not available? My suggestion would be: error mesage and quit.
xembedsniproxy/fdoselectionmanager.cpp (line 64)
<https://git.reviewboard.kde.org/r/125655/#comment59771>
qCDebug?
xembedsniproxy/main.cpp (line 35)
<https://git.reviewboard.kde.org/r/125655/#comment59774>
please enforce platform XCB, you are using X11 unconditionally.
xembedsniproxy/main.cpp (line 43)
<https://git.reviewboard.kde.org/r/125655/#comment59775>
QScopedPointer
xembedsniproxy/sniproxy.h (lines 42 - 50)
<https://git.reviewboard.kde.org/r/125655/#comment59776>
hmm?
xembedsniproxy/sniproxy.cpp (line 119)
<https://git.reviewboard.kde.org/r/125655/#comment59777>
please do even one more thing (if possible): render once to it with a translucent background. The tricky part is to get the correct depth for it.
xembedsniproxy/sniproxy.cpp (lines 180 - 183)
<https://git.reviewboard.kde.org/r/125655/#comment59778>
maybe you could use the sync protocol?
xembedsniproxy/sniproxy.cpp (line 218)
<https://git.reviewboard.kde.org/r/125655/#comment59779>
no cast from ascii please ;-)
xembedsniproxy/sniproxy.cpp (line 239)
<https://git.reviewboard.kde.org/r/125655/#comment59780>
same here: no cast from ascii
- Martin Gräßlin
On Oct. 16, 2015, 1:22 p.m., David Edmundson wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://git.reviewboard.kde.org/r/125655/
> -----------------------------------------------------------
>
> (Updated Oct. 16, 2015, 1:22 p.m.)
>
>
> Review request for Plasma.
>
>
> Repository: plasma-workspace
>
>
> Description
> -------
>
> The goal of this project is to make xembed system trays available in Plasma.
>
> This is to allow legacy apps (xchat, pidgin, tuxguitar) etc. system trays[1] available in Plasma which only supports StatusNotifierItem [2].
>
> Ideally we also want this to work in an xwayland session, making X system tray icons available even when plasmashell only has a wayland connection.
>
> How it works (in theory)
>
> * We register a window as a system tray container
> * We render embeded windows composited offscreen
> * We render contents into an image and send this over DBus via the SNI protocol
> * XDamage events trigger a repaint
> * Activate and context menu events are replyed via X send event into the embedded container as left and right clicks
>
>
> Code is a mix of hacks and bodges bashing client apps into place, and I happily acknowledge it looks pretty bad. Unfotunately most of it seems to be needed, but I'd love to be proved wrong on most of it.
>
> ---
>
> CMakeLists changes are obviously currently wrong, I can strip half of find_packages and I need an add_subdirectory in the root file, but it's easier for me to sync with the separate repo mirror this way till we're close to merging. I expect this to take quite some revisions.
>
>
> Diffs
> -----
>
> xembedsniproxy/CMakeLists.txt PRE-CREATION
> xembedsniproxy/Readme.md PRE-CREATION
> xembedsniproxy/fdoselectionmanager.h PRE-CREATION
> xembedsniproxy/fdoselectionmanager.cpp PRE-CREATION
> xembedsniproxy/main.cpp PRE-CREATION
> xembedsniproxy/org.kde.StatusNotifierItem.xml PRE-CREATION
> xembedsniproxy/org.kde.StatusNotifierWatcher.xml PRE-CREATION
> xembedsniproxy/snidbus.h PRE-CREATION
> xembedsniproxy/snidbus.cpp PRE-CREATION
> xembedsniproxy/sniproxy.h PRE-CREATION
> xembedsniproxy/sniproxy.cpp PRE-CREATION
> xembedsniproxy/xcbutils.h PRE-CREATION
> xembedsniproxy/xembedsniproxy.desktop PRE-CREATION
>
> Diff: https://git.reviewboard.kde.org/r/125655/diff/
>
>
> Testing
> -------
>
>
> Thanks,
>
> David Edmundson
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/plasma-devel/attachments/20151016/d6eb5aad/attachment-0001.html>
More information about the Plasma-devel
mailing list