From kde-commits  Thu Jan 08 23:42:40 2009
From: "Aaron J. Seigo" <aseigo () kde ! org>
Date: Thu, 08 Jan 2009 23:42:40 +0000
To: kde-commits
Subject: Re: KDE/kdebase/workspace/plasma/applets/systemtray/protocols/fdo
Message-Id: <200901081741.34460.aseigo () kde ! org>
X-MARC-Message: https://marc.info/?l=kde-commits&m=123145816014375
MIME-Version: 1
Content-Type: multipart/mixed; boundary="--nextPart3805020.RQ1LWFGfax"

--nextPart3805020.RQ1LWFGfax
Content-Type: text/plain;
  charset="utf-8"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

On Thursday 08 January 2009, Fredrik H=C3=B6glund wrote:
> On Thursday 08 January 2009 18:59, Aaron J. Seigo wrote:
> > SVN commit 907753 by aseigo:
> >
> > don't assume 32 bit depth, makes trays work on 16 bit depth with the new
> > hint. patch by ddenis @ Qt; thanks! =3D)
> >
> >
> >  M  +16 -12    fdoselectionmanager.cpp
> >  M  +0 -4      x11embedcontainer.cpp
> >
> >
> > ---
> > trunk/KDE/kdebase/workspace/plasma/applets/systemtray/protocols/fdo/fdo=
se
> >lectionmanager.cpp #907752:907753 @@ -267,20 +267,24 @@
> >      int nvi;
> >      VisualID visual =3D
> > XVisualIDFromVisual((Visual*)QX11Info::appVisual()); XVisualInfo templ;
> > -    templ.screen  =3D DefaultScreen(d->display);
> > -    templ.depth   =3D 32;
> > -    templ.c_class =3D TrueColor;
> > -    XVisualInfo *xvi =3D XGetVisualInfo(d->display, VisualScreenMask |
> > VisualDepthMask | VisualClassMask, -                                   =
 =20
> > &templ, &nvi);
> > -    for (int i =3D 0; i < nvi; i++) {
> > -        XRenderPictFormat *format =3D XRenderFindVisualFormat(d->displ=
ay,
> > xvi[i].visual); -        if (format->type =3D=3D PictTypeDirect &&
> > format->direct.alphaMask) { -            visual =3D xvi[i].visualid;
> > -            break;
> > +    templ.visualid =3D visual;
> > +    XVisualInfo *xvi =3D XGetVisualInfo(d->display, VisualIDMask, &tem=
pl,
> > &nvi); +    if (xvi) {
> > +        templ.screen  =3D xvi[0].screen;
> > +        templ.depth   =3D xvi[0].depth;
> > +        templ.c_class =3D xvi[0].c_class;
> > +        XFree(xvi);
> > +        xvi =3D XGetVisualInfo(d->display, VisualScreenMask |
> > VisualDepthMask | VisualClassMask, +                             &templ,
> > &nvi);
> > +        for (int i =3D 0; i < nvi; i++) {
> > +            XRenderPictFormat *format =3D
> > XRenderFindVisualFormat(d->display, xvi[i].visual); +            if
> > (format->type =3D=3D PictTypeDirect && format->direct.alphaMask) { +   =
     =20
> >       visual =3D xvi[i].visualid;
> > +                break;
> > +            }
> >          }
> > +        XFree(xvi);
> >      }
>
> You just turned this whole block of code into an elaborate no-op that
> always ends up right back where you started; with the visual Plasma was
> using in the first place.
>
> The whole point of this code is to tell applications that you want systray
> icons to always use the 32 bit ARGB visual regardless of the display dept=
h,
> so they will always have an alpha channel.
>
> The new code still accomplishes that by accident, but only because Plasma
> itself always uses the ARGB visual. But the minute you remove that hack
> and start using the new Qt::WA_TranslucentBackground attribute, you
> break the systray.
>
> So could you explain what you were trying to accomplish with this patch
> so we can fix it properly?

according to ddenis who worked on the NET_SYSTEM_TRAY_VISUAL for Qt 4.5, th=
e=20
old code was failing on 16bpp displays. well, i'll let him speak for himsel=
f=20
here:

[Thu Jan 8 2009] [03:55:29] <ddenis> Hi. I wanted to talk to about about th=
e=20
NET_SYSTEM_TRAY_VISUAL atom - I've implemented support for it in Qt 4.5,=20
however I suspect that its implementation in the systemtray applet in kde i=
s=20
not correct - in the fdo/fdoselectionmanager.cpp in the initSelection()=20
function the code that searches for the visual explicitely sets the color=20
depth to be 32 and the problem occurs when I try to create a systray icon=20
window with that visual on a 16bit display. As far as
[Thu Jan 8 2009] [03:55:29] <ddenis> I can see the proper way is to get the=
=20
depth from the system and then try to find the ARGB-enabled visual with the=
=20
same depth on the same screen

he provided a patch that he claimed worked for him in those situations, i=20
applied and tested it here as well ... but perhaps you can propose a better=
=20
solution that isn't an elaborate no-op? ;)

=2D-=20
Aaron J. Seigo
humru othro a kohnu se
GPG Fingerprint: 8B8B 2209 0C6F 7C47 B1EA  EE75 D6B7 2EB1 A7F1 DB43

KDE core developer sponsored by Qt Software


--nextPart3805020.RQ1LWFGfax
Content-Type: application/pgp-signature; name=signature.asc 
Content-Description: This is a digitally signed message part.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.9 (GNU/Linux)

iEYEABECAAYFAklmnT4ACgkQ1rcusafx20OM7wCgmZkbi3cEpUqbHL89/REpU6E+
JVMAn28tqHCRsiHgQt7HALVbjOokwT/r
=sHKm
-----END PGP SIGNATURE-----

--nextPart3805020.RQ1LWFGfax--