| Submitter | Otavio Salvador |
|---|---|
| Date | Feb. 16, 2013, 9:20 p.m. |
| Message ID | <1361049644-27904-2-git-send-email-otavio@ossystems.com.br> |
| Download | mbox | patch |
| Permalink | /patch/44723/ |
| State | New |
| Headers | show |
Comments
On Sat, 2013-02-16 at 19:20 -0200, Otavio Salvador wrote: > Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> > --- > meta/recipes-core/dbus/dbus.inc | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/meta/recipes-core/dbus/dbus.inc b/meta/recipes-core/dbus/dbus.inc > index 292ab29..cebab28 100644 > --- a/meta/recipes-core/dbus/dbus.inc > +++ b/meta/recipes-core/dbus/dbus.inc > @@ -8,7 +8,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=10dded3b58148f3f1fd804b26354af3e \ > X11DEPENDS = "virtual/libx11 libsm" > DEPENDS = "expat virtual/libintl ${@base_contains('DISTRO_FEATURES', 'x11', '${X11DEPENDS}', '', d)}" > DEPENDS_class-native = "expat-native virtual/libintl-native" > -DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl virtual/libx11" > +DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl ${@base_contains('DISTRO_FEATURES', 'x11', 'virtual/libx11', '', d)}" > RDEPENDS_dbus = "${@base_contains('DISTRO_FEATURES', 'ptest', 'dbus-ptest', '', d)}" > RDEPENDS_class-native = "" This tends to be a really bad idea for dbus-nativesdk. If you install an SDK compiled with these options into PATH only a system which does have x11, it will break functionality in a way unacceptable to users. I'm not sure we should ever ship dbus-launch with the SDK so perhaps we should just delete it like we now do for the -native version, then we could drop the x11 requirement in all cases. Cheers, Richard
On Sun, Feb 17, 2013 at 4:47 AM, Richard Purdie <richard.purdie@linuxfoundation.org> wrote: > On Sat, 2013-02-16 at 19:20 -0200, Otavio Salvador wrote: >> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> >> --- >> meta/recipes-core/dbus/dbus.inc | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/meta/recipes-core/dbus/dbus.inc b/meta/recipes-core/dbus/dbus.inc >> index 292ab29..cebab28 100644 >> --- a/meta/recipes-core/dbus/dbus.inc >> +++ b/meta/recipes-core/dbus/dbus.inc >> @@ -8,7 +8,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=10dded3b58148f3f1fd804b26354af3e \ >> X11DEPENDS = "virtual/libx11 libsm" >> DEPENDS = "expat virtual/libintl ${@base_contains('DISTRO_FEATURES', 'x11', '${X11DEPENDS}', '', d)}" >> DEPENDS_class-native = "expat-native virtual/libintl-native" >> -DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl virtual/libx11" >> +DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl ${@base_contains('DISTRO_FEATURES', 'x11', 'virtual/libx11', '', d)}" >> RDEPENDS_dbus = "${@base_contains('DISTRO_FEATURES', 'ptest', 'dbus-ptest', '', d)}" >> RDEPENDS_class-native = "" > > This tends to be a really bad idea for dbus-nativesdk. If you install an > SDK compiled with these options into PATH only a system which does have > x11, it will break functionality in a way unacceptable to users. I don't see how it'd break anything being deployed in a toolchain. We don't initialize the dbus daemon so it is mostly used to fulfill dependencies. > I'm not sure we should ever ship dbus-launch with the SDK so perhaps we > should just delete it like we now do for the -native version, then we > could drop the x11 requirement in all cases. That could be an option (ignore x11 feature for nativesdk) and it does make some sense.
On Sun, 2013-02-17 at 10:40 -0300, Otavio Salvador wrote: > On Sun, Feb 17, 2013 at 4:47 AM, Richard Purdie > <richard.purdie@linuxfoundation.org> wrote: > > On Sat, 2013-02-16 at 19:20 -0200, Otavio Salvador wrote: > >> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> > >> --- > >> meta/recipes-core/dbus/dbus.inc | 2 +- > >> 1 file changed, 1 insertion(+), 1 deletion(-) > >> > >> diff --git a/meta/recipes-core/dbus/dbus.inc b/meta/recipes-core/dbus/dbus.inc > >> index 292ab29..cebab28 100644 > >> --- a/meta/recipes-core/dbus/dbus.inc > >> +++ b/meta/recipes-core/dbus/dbus.inc > >> @@ -8,7 +8,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=10dded3b58148f3f1fd804b26354af3e \ > >> X11DEPENDS = "virtual/libx11 libsm" > >> DEPENDS = "expat virtual/libintl ${@base_contains('DISTRO_FEATURES', 'x11', '${X11DEPENDS}', '', d)}" > >> DEPENDS_class-native = "expat-native virtual/libintl-native" > >> -DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl virtual/libx11" > >> +DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl ${@base_contains('DISTRO_FEATURES', 'x11', 'virtual/libx11', '', d)}" > >> RDEPENDS_dbus = "${@base_contains('DISTRO_FEATURES', 'ptest', 'dbus-ptest', '', d)}" > >> RDEPENDS_class-native = "" > > > > This tends to be a really bad idea for dbus-nativesdk. If you install an > > SDK compiled with these options into PATH only a system which does have > > x11, it will break functionality in a way unacceptable to users. > > I don't see how it'd break anything being deployed in a toolchain. We > don't initialize the dbus daemon so it is mostly used to fulfill > dependencies. Imagine you source the toolchain setup script. The limited functionality dbus-launch just got into PATH. Now you run some X application. Things break. Cheers, Richard
On Sun, Feb 17, 2013 at 10:44 AM, Richard Purdie <richard.purdie@linuxfoundation.org> wrote: > On Sun, 2013-02-17 at 10:40 -0300, Otavio Salvador wrote: >> On Sun, Feb 17, 2013 at 4:47 AM, Richard Purdie >> <richard.purdie@linuxfoundation.org> wrote: >> > On Sat, 2013-02-16 at 19:20 -0200, Otavio Salvador wrote: >> >> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> >> >> --- >> >> meta/recipes-core/dbus/dbus.inc | 2 +- >> >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> >> >> diff --git a/meta/recipes-core/dbus/dbus.inc b/meta/recipes-core/dbus/dbus.inc >> >> index 292ab29..cebab28 100644 >> >> --- a/meta/recipes-core/dbus/dbus.inc >> >> +++ b/meta/recipes-core/dbus/dbus.inc >> >> @@ -8,7 +8,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=10dded3b58148f3f1fd804b26354af3e \ >> >> X11DEPENDS = "virtual/libx11 libsm" >> >> DEPENDS = "expat virtual/libintl ${@base_contains('DISTRO_FEATURES', 'x11', '${X11DEPENDS}', '', d)}" >> >> DEPENDS_class-native = "expat-native virtual/libintl-native" >> >> -DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl virtual/libx11" >> >> +DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl ${@base_contains('DISTRO_FEATURES', 'x11', 'virtual/libx11', '', d)}" >> >> RDEPENDS_dbus = "${@base_contains('DISTRO_FEATURES', 'ptest', 'dbus-ptest', '', d)}" >> >> RDEPENDS_class-native = "" >> > >> > This tends to be a really bad idea for dbus-nativesdk. If you install an >> > SDK compiled with these options into PATH only a system which does have >> > x11, it will break functionality in a way unacceptable to users. >> >> I don't see how it'd break anything being deployed in a toolchain. We >> don't initialize the dbus daemon so it is mostly used to fulfill >> dependencies. > > Imagine you source the toolchain setup script. The limited functionality > dbus-launch just got into PATH. Now you run some X application. Things > break. You usually to not source the env script and run your X server on top of it, I think. Personally I source it in a terminal session inside of X and use it to build things only.
On Sun, 2013-02-17 at 10:55 -0300, Otavio Salvador wrote: > On Sun, Feb 17, 2013 at 10:44 AM, Richard Purdie > <richard.purdie@linuxfoundation.org> wrote: > > On Sun, 2013-02-17 at 10:40 -0300, Otavio Salvador wrote: > >> On Sun, Feb 17, 2013 at 4:47 AM, Richard Purdie > >> <richard.purdie@linuxfoundation.org> wrote: > >> > On Sat, 2013-02-16 at 19:20 -0200, Otavio Salvador wrote: > >> >> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> > >> >> --- > >> >> meta/recipes-core/dbus/dbus.inc | 2 +- > >> >> 1 file changed, 1 insertion(+), 1 deletion(-) > >> >> > >> >> diff --git a/meta/recipes-core/dbus/dbus.inc b/meta/recipes-core/dbus/dbus.inc > >> >> index 292ab29..cebab28 100644 > >> >> --- a/meta/recipes-core/dbus/dbus.inc > >> >> +++ b/meta/recipes-core/dbus/dbus.inc > >> >> @@ -8,7 +8,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=10dded3b58148f3f1fd804b26354af3e \ > >> >> X11DEPENDS = "virtual/libx11 libsm" > >> >> DEPENDS = "expat virtual/libintl ${@base_contains('DISTRO_FEATURES', 'x11', '${X11DEPENDS}', '', d)}" > >> >> DEPENDS_class-native = "expat-native virtual/libintl-native" > >> >> -DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl virtual/libx11" > >> >> +DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl ${@base_contains('DISTRO_FEATURES', 'x11', 'virtual/libx11', '', d)}" > >> >> RDEPENDS_dbus = "${@base_contains('DISTRO_FEATURES', 'ptest', 'dbus-ptest', '', d)}" > >> >> RDEPENDS_class-native = "" > >> > > >> > This tends to be a really bad idea for dbus-nativesdk. If you install an > >> > SDK compiled with these options into PATH only a system which does have > >> > x11, it will break functionality in a way unacceptable to users. > >> > >> I don't see how it'd break anything being deployed in a toolchain. We > >> don't initialize the dbus daemon so it is mostly used to fulfill > >> dependencies. > > > > Imagine you source the toolchain setup script. The limited functionality > > dbus-launch just got into PATH. Now you run some X application. Things > > break. > > You usually to not source the env script and run your X server on top > of it, I think. Personally I source it in a terminal session inside of > X and use it to build things only. No, but people do open a terminal in X and then source the script there, then run other X applications from within the terminal window. Those X applications can use dbus activation. A dbus-launch without x11 support causes problems in this situation. Cheers, Richard
On Sun, Feb 17, 2013 at 11:48 AM, Richard Purdie <richard.purdie@linuxfoundation.org> wrote: > On Sun, 2013-02-17 at 10:55 -0300, Otavio Salvador wrote: >> On Sun, Feb 17, 2013 at 10:44 AM, Richard Purdie >> <richard.purdie@linuxfoundation.org> wrote: >> > On Sun, 2013-02-17 at 10:40 -0300, Otavio Salvador wrote: >> >> On Sun, Feb 17, 2013 at 4:47 AM, Richard Purdie >> >> <richard.purdie@linuxfoundation.org> wrote: >> >> > On Sat, 2013-02-16 at 19:20 -0200, Otavio Salvador wrote: >> >> >> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> >> >> >> --- >> >> >> meta/recipes-core/dbus/dbus.inc | 2 +- >> >> >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> >> >> >> >> diff --git a/meta/recipes-core/dbus/dbus.inc b/meta/recipes-core/dbus/dbus.inc >> >> >> index 292ab29..cebab28 100644 >> >> >> --- a/meta/recipes-core/dbus/dbus.inc >> >> >> +++ b/meta/recipes-core/dbus/dbus.inc >> >> >> @@ -8,7 +8,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=10dded3b58148f3f1fd804b26354af3e \ >> >> >> X11DEPENDS = "virtual/libx11 libsm" >> >> >> DEPENDS = "expat virtual/libintl ${@base_contains('DISTRO_FEATURES', 'x11', '${X11DEPENDS}', '', d)}" >> >> >> DEPENDS_class-native = "expat-native virtual/libintl-native" >> >> >> -DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl virtual/libx11" >> >> >> +DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl ${@base_contains('DISTRO_FEATURES', 'x11', 'virtual/libx11', '', d)}" >> >> >> RDEPENDS_dbus = "${@base_contains('DISTRO_FEATURES', 'ptest', 'dbus-ptest', '', d)}" >> >> >> RDEPENDS_class-native = "" >> >> > >> >> > This tends to be a really bad idea for dbus-nativesdk. If you install an >> >> > SDK compiled with these options into PATH only a system which does have >> >> > x11, it will break functionality in a way unacceptable to users. >> >> >> >> I don't see how it'd break anything being deployed in a toolchain. We >> >> don't initialize the dbus daemon so it is mostly used to fulfill >> >> dependencies. >> > >> > Imagine you source the toolchain setup script. The limited functionality >> > dbus-launch just got into PATH. Now you run some X application. Things >> > break. >> >> You usually to not source the env script and run your X server on top >> of it, I think. Personally I source it in a terminal session inside of >> X and use it to build things only. > > No, but people do open a terminal in X and then source the script there, > then run other X applications from within the terminal window. Those X > applications can use dbus activation. A dbus-launch without x11 support > causes problems in this situation. Yes; got it. So your idea is to avoid x11 dependency altogether and drop dbus-launch to avoid this errors?
On Sun, 2013-02-17 at 11:52 -0300, Otavio Salvador wrote: > On Sun, Feb 17, 2013 at 11:48 AM, Richard Purdie > <richard.purdie@linuxfoundation.org> wrote: > > On Sun, 2013-02-17 at 10:55 -0300, Otavio Salvador wrote: > >> On Sun, Feb 17, 2013 at 10:44 AM, Richard Purdie > >> <richard.purdie@linuxfoundation.org> wrote: > >> > On Sun, 2013-02-17 at 10:40 -0300, Otavio Salvador wrote: > >> >> On Sun, Feb 17, 2013 at 4:47 AM, Richard Purdie > >> >> <richard.purdie@linuxfoundation.org> wrote: > >> >> > On Sat, 2013-02-16 at 19:20 -0200, Otavio Salvador wrote: > >> >> >> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> > >> >> >> --- > >> >> >> meta/recipes-core/dbus/dbus.inc | 2 +- > >> >> >> 1 file changed, 1 insertion(+), 1 deletion(-) > >> >> >> > >> >> >> diff --git a/meta/recipes-core/dbus/dbus.inc b/meta/recipes-core/dbus/dbus.inc > >> >> >> index 292ab29..cebab28 100644 > >> >> >> --- a/meta/recipes-core/dbus/dbus.inc > >> >> >> +++ b/meta/recipes-core/dbus/dbus.inc > >> >> >> @@ -8,7 +8,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=10dded3b58148f3f1fd804b26354af3e \ > >> >> >> X11DEPENDS = "virtual/libx11 libsm" > >> >> >> DEPENDS = "expat virtual/libintl ${@base_contains('DISTRO_FEATURES', 'x11', '${X11DEPENDS}', '', d)}" > >> >> >> DEPENDS_class-native = "expat-native virtual/libintl-native" > >> >> >> -DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl virtual/libx11" > >> >> >> +DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl ${@base_contains('DISTRO_FEATURES', 'x11', 'virtual/libx11', '', d)}" > >> >> >> RDEPENDS_dbus = "${@base_contains('DISTRO_FEATURES', 'ptest', 'dbus-ptest', '', d)}" > >> >> >> RDEPENDS_class-native = "" > >> >> > > >> >> > This tends to be a really bad idea for dbus-nativesdk. If you install an > >> >> > SDK compiled with these options into PATH only a system which does have > >> >> > x11, it will break functionality in a way unacceptable to users. > >> >> > >> >> I don't see how it'd break anything being deployed in a toolchain. We > >> >> don't initialize the dbus daemon so it is mostly used to fulfill > >> >> dependencies. > >> > > >> > Imagine you source the toolchain setup script. The limited functionality > >> > dbus-launch just got into PATH. Now you run some X application. Things > >> > break. > >> > >> You usually to not source the env script and run your X server on top > >> of it, I think. Personally I source it in a terminal session inside of > >> X and use it to build things only. > > > > No, but people do open a terminal in X and then source the script there, > > then run other X applications from within the terminal window. Those X > > applications can use dbus activation. A dbus-launch without x11 support > > causes problems in this situation. > > Yes; got it. > > So your idea is to avoid x11 dependency altogether and drop > dbus-launch to avoid this errors? Yes, I did the same thing with dbus-native for exactly this reason (e.g. X apps from within a devshell). I didn't touch dbus-nativesdk since it has the X dependency and I didn't think it was bothering anyone. Cheers, Richard
On Sun, Feb 17, 2013 at 12:10 PM, Richard Purdie <richard.purdie@linuxfoundation.org> wrote: > On Sun, 2013-02-17 at 11:52 -0300, Otavio Salvador wrote: >> On Sun, Feb 17, 2013 at 11:48 AM, Richard Purdie >> <richard.purdie@linuxfoundation.org> wrote: >> > On Sun, 2013-02-17 at 10:55 -0300, Otavio Salvador wrote: >> >> On Sun, Feb 17, 2013 at 10:44 AM, Richard Purdie >> >> <richard.purdie@linuxfoundation.org> wrote: >> >> > On Sun, 2013-02-17 at 10:40 -0300, Otavio Salvador wrote: >> >> >> On Sun, Feb 17, 2013 at 4:47 AM, Richard Purdie >> >> >> <richard.purdie@linuxfoundation.org> wrote: >> >> >> > On Sat, 2013-02-16 at 19:20 -0200, Otavio Salvador wrote: >> >> >> >> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> >> >> >> >> --- >> >> >> >> meta/recipes-core/dbus/dbus.inc | 2 +- >> >> >> >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> >> >> >> >> >> >> diff --git a/meta/recipes-core/dbus/dbus.inc b/meta/recipes-core/dbus/dbus.inc >> >> >> >> index 292ab29..cebab28 100644 >> >> >> >> --- a/meta/recipes-core/dbus/dbus.inc >> >> >> >> +++ b/meta/recipes-core/dbus/dbus.inc >> >> >> >> @@ -8,7 +8,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=10dded3b58148f3f1fd804b26354af3e \ >> >> >> >> X11DEPENDS = "virtual/libx11 libsm" >> >> >> >> DEPENDS = "expat virtual/libintl ${@base_contains('DISTRO_FEATURES', 'x11', '${X11DEPENDS}', '', d)}" >> >> >> >> DEPENDS_class-native = "expat-native virtual/libintl-native" >> >> >> >> -DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl virtual/libx11" >> >> >> >> +DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl ${@base_contains('DISTRO_FEATURES', 'x11', 'virtual/libx11', '', d)}" >> >> >> >> RDEPENDS_dbus = "${@base_contains('DISTRO_FEATURES', 'ptest', 'dbus-ptest', '', d)}" >> >> >> >> RDEPENDS_class-native = "" >> >> >> > >> >> >> > This tends to be a really bad idea for dbus-nativesdk. If you install an >> >> >> > SDK compiled with these options into PATH only a system which does have >> >> >> > x11, it will break functionality in a way unacceptable to users. >> >> >> >> >> >> I don't see how it'd break anything being deployed in a toolchain. We >> >> >> don't initialize the dbus daemon so it is mostly used to fulfill >> >> >> dependencies. >> >> > >> >> > Imagine you source the toolchain setup script. The limited functionality >> >> > dbus-launch just got into PATH. Now you run some X application. Things >> >> > break. >> >> >> >> You usually to not source the env script and run your X server on top >> >> of it, I think. Personally I source it in a terminal session inside of >> >> X and use it to build things only. >> > >> > No, but people do open a terminal in X and then source the script there, >> > then run other X applications from within the terminal window. Those X >> > applications can use dbus activation. A dbus-launch without x11 support >> > causes problems in this situation. >> >> Yes; got it. >> >> So your idea is to avoid x11 dependency altogether and drop >> dbus-launch to avoid this errors? > > Yes, I did the same thing with dbus-native for exactly this reason (e.g. > X apps from within a devshell). I didn't touch dbus-nativesdk since it > has the X dependency and I didn't think it was bothering anyone. Ok; will make v2 with this change.
Patch
diff --git a/meta/recipes-core/dbus/dbus.inc b/meta/recipes-core/dbus/dbus.inc index 292ab29..cebab28 100644 --- a/meta/recipes-core/dbus/dbus.inc +++ b/meta/recipes-core/dbus/dbus.inc @@ -8,7 +8,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=10dded3b58148f3f1fd804b26354af3e \ X11DEPENDS = "virtual/libx11 libsm" DEPENDS = "expat virtual/libintl ${@base_contains('DISTRO_FEATURES', 'x11', '${X11DEPENDS}', '', d)}" DEPENDS_class-native = "expat-native virtual/libintl-native" -DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl virtual/libx11" +DEPENDS_class-nativesdk = "nativesdk-expat virtual/nativesdk-libintl ${@base_contains('DISTRO_FEATURES', 'x11', 'virtual/libx11', '', d)}" RDEPENDS_dbus = "${@base_contains('DISTRO_FEATURES', 'ptest', 'dbus-ptest', '', d)}" RDEPENDS_class-native = ""
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> --- meta/recipes-core/dbus/dbus.inc | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)