virtclass_install prepend

virtclass_install prepend

[Frans Meulenbroeks]

Hi,

Found this in our reicpes:

dbus/dbus-glib_0.86.bb:do_install_virtclass-native_prepend() {
postfix/postfix.inc:do_install_prepend_virtclass-native() {

Who is right wrt the position of _prepend ?

Or are both ok ? (and if so are there any differences? )

Frans

Re: virtclass_install prepend

[Chris Larson]

On Thu, Aug 19, 2010 at 4:29 AM, Frans Meulenbroeks <
fransmeulenbroeks@gmail.com> wrote:

> Hi,
>
> Found this in our reicpes:
>
> dbus/dbus-glib_0.86.bb:do_install_virtclass-native_prepend() {
>

This prepends to do_install_virtclass-native, which will then replace
do_install when the overrides are applied.  This is prepending to a variable
which overrides, as opposed to the other, which is a conditional prepend.
 The two are quite different.  It's unlikely that this is what they
intended.  Far more likely that it should be changed to
_prepend_virtclass-native.
-- 
Christopher Larson
clarson at kergoth dot com
Founder - BitBake, OpenEmbedded, OpenZaurus
Maintainer - Tslib
Senior Software Engineer, Mentor Graphics

Re: virtclass_install prepend

[Frans Meulenbroeks]

2010/8/19 Chris Larson <clarson@kergoth.com>:
> On Thu, Aug 19, 2010 at 4:29 AM, Frans Meulenbroeks <
> fransmeulenbroeks@gmail.com> wrote:
>
>> Hi,
>>
>> Found this in our reicpes:
>>
>> dbus/dbus-glib_0.86.bb:do_install_virtclass-native_prepend() {
>>
>
> This prepends to do_install_virtclass-native, which will then replace
> do_install when the overrides are applied.  This is prepending to a variable
> which overrides, as opposed to the other, which is a conditional prepend.
>  The two are quite different.  It's unlikely that this is what they
> intended.  Far more likely that it should be changed to
> _prepend_virtclass-native.

I have no idea what is intended and I have no dbus knowledge, so can't fix this.
I'll leave it to someone else to resolve this.

Frans

Re: virtclass_install prepend

[Khem Raj]

On (19/08/10 21:14), Frans Meulenbroeks wrote:
> 2010/8/19 Chris Larson <clarson@kergoth.com>:
> > On Thu, Aug 19, 2010 at 4:29 AM, Frans Meulenbroeks <
> > fransmeulenbroeks@gmail.com> wrote:
> >
> >> Hi,
> >>
> >> Found this in our reicpes:
> >>
> >> dbus/dbus-glib_0.86.bb:do_install_virtclass-native_prepend() {
> >>
> >
> > This prepends to do_install_virtclass-native, which will then replace
> > do_install when the overrides are applied.  This is prepending to a variable
> > which overrides, as opposed to the other, which is a conditional prepend.
> >  The two are quite different.  It's unlikely that this is what they
> > intended.  Far more likely that it should be changed to
> > _prepend_virtclass-native.
> 
> I have no idea what is intended and I have no dbus knowledge, so can't fix this.
> I'll leave it to someone else to resolve this.

I think the correct fix is as Chris suggested. Please go ahead and fix it.

> 
> Frans
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

Re: virtclass_install prepend

[Frans Meulenbroeks]

2010/9/6 Khem Raj <raj.khem@gmail.com>:
> On (19/08/10 21:14), Frans Meulenbroeks wrote:
>> 2010/8/19 Chris Larson <clarson@kergoth.com>:
>> > On Thu, Aug 19, 2010 at 4:29 AM, Frans Meulenbroeks <
>> > fransmeulenbroeks@gmail.com> wrote:
>> >
>> >> Hi,
>> >>
>> >> Found this in our reicpes:
>> >>
>> >> dbus/dbus-glib_0.86.bb:do_install_virtclass-native_prepend() {
>> >>
>> >
>> > This prepends to do_install_virtclass-native, which will then replace
>> > do_install when the overrides are applied.  This is prepending to a variable
>> > which overrides, as opposed to the other, which is a conditional prepend.
>> >  The two are quite different.  It's unlikely that this is what they
>> > intended.  Far more likely that it should be changed to
>> > _prepend_virtclass-native.
>>
>> I have no idea what is intended and I have no dbus knowledge, so can't fix this.
>> I'll leave it to someone else to resolve this.
>
> I think the correct fix is as Chris suggested. Please go ahead and fix it.
>

Well I am a little bit hesitating as I am by no means an expert on the
dbus recipe.

The other question that immediately popped up is:
If the code is wrong and no-one noticed, is it actually needed?

The code is:
do_install_virtclass-native_prepend() {
        install -d ${D}${datadir}/dbus
        install -m 0644 dbus-bus-introspect.xml ${D}${datadir}/dbus
}

Actually if the standard do_install needs to be changed, it generally
raises some alarms with me.
Why would we need to install something that the package normally does
not install?
(of course this could be a bug in the package, but then I would prefer
a patch that we also try to push upstream).

Frans.

Re: virtclass_install prepend

[Chris Larson]

On Mon, Sep 6, 2010 at 12:41 AM, Frans Meulenbroeks <
fransmeulenbroeks@gmail.com> wrote:

> 2010/9/6 Khem Raj <raj.khem@gmail.com>:
> > On (19/08/10 21:14), Frans Meulenbroeks wrote:
> >> 2010/8/19 Chris Larson <clarson@kergoth.com>:
> >> > On Thu, Aug 19, 2010 at 4:29 AM, Frans Meulenbroeks <
> >> > fransmeulenbroeks@gmail.com> wrote:
> >> >
> >> >> Hi,
> >> >>
> >> >> Found this in our reicpes:
> >> >>
> >> >> dbus/dbus-glib_0.86.bb:do_install_virtclass-native_prepend() {
> >> >>
> >> >
> >> > This prepends to do_install_virtclass-native, which will then replace
> >> > do_install when the overrides are applied.  This is prepending to a
> variable
> >> > which overrides, as opposed to the other, which is a conditional
> prepend.
> >> >  The two are quite different.  It's unlikely that this is what they
> >> > intended.  Far more likely that it should be changed to
> >> > _prepend_virtclass-native.
> >>
> >> I have no idea what is intended and I have no dbus knowledge, so can't
> fix this.
> >> I'll leave it to someone else to resolve this.
> >
> > I think the correct fix is as Chris suggested. Please go ahead and fix
> it.
> >
>
> Well I am a little bit hesitating as I am by no means an expert on the
> dbus recipe.
>
> The other question that immediately popped up is:
> If the code is wrong and no-one noticed, is it actually needed?
>
> The code is:
> do_install_virtclass-native_prepend() {
>         install -d ${D}${datadir}/dbus
>        install -m 0644 dbus-bus-introspect.xml ${D}${datadir}/dbus
> }
>
> Actually if the standard do_install needs to be changed, it generally
> raises some alarms with me.
> Why would we need to install something that the package normally does
> not install?
> (of course this could be a bug in the package, but then I would prefer
> a patch that we also try to push upstream).
>

Agreed, that is odd.  If it's something that dbus using things might need,
why is it only prepended for the native version, and as you say, why is it
not installed by the upstream install target.  I'd do a git blame and figure
out who added that and ask them :)
-- 
Christopher Larson
clarson at kergoth dot com
Founder - BitBake, OpenEmbedded, OpenZaurus
Maintainer - Tslib
Senior Software Engineer, Mentor Graphics

Re: virtclass_install prepend

[Khem Raj]

On Mon, Sep 6, 2010 at 6:54 AM, Chris Larson <clarson@kergoth.com> wrote:
> On Mon, Sep 6, 2010 at 12:41 AM, Frans Meulenbroeks <
> fransmeulenbroeks@gmail.com> wrote:
>
>> 2010/9/6 Khem Raj <raj.khem@gmail.com>:
>> > On (19/08/10 21:14), Frans Meulenbroeks wrote:
>> >> 2010/8/19 Chris Larson <clarson@kergoth.com>:
>> >> > On Thu, Aug 19, 2010 at 4:29 AM, Frans Meulenbroeks <
>> >> > fransmeulenbroeks@gmail.com> wrote:
>> >> >
>> >> >> Hi,
>> >> >>
>> >> >> Found this in our reicpes:
>> >> >>
>> >> >> dbus/dbus-glib_0.86.bb:do_install_virtclass-native_prepend() {
>> >> >>
>> >> >
>> >> > This prepends to do_install_virtclass-native, which will then replace
>> >> > do_install when the overrides are applied.  This is prepending to a
>> variable
>> >> > which overrides, as opposed to the other, which is a conditional
>> prepend.
>> >> >  The two are quite different.  It's unlikely that this is what they
>> >> > intended.  Far more likely that it should be changed to
>> >> > _prepend_virtclass-native.
>> >>
>> >> I have no idea what is intended and I have no dbus knowledge, so can't
>> fix this.
>> >> I'll leave it to someone else to resolve this.
>> >
>> > I think the correct fix is as Chris suggested. Please go ahead and fix
>> it.
>> >
>>
>> Well I am a little bit hesitating as I am by no means an expert on the
>> dbus recipe.
>>
>> The other question that immediately popped up is:
>> If the code is wrong and no-one noticed, is it actually needed?
>>
>> The code is:
>> do_install_virtclass-native_prepend() {
>>         install -d ${D}${datadir}/dbus
>>        install -m 0644 dbus-bus-introspect.xml ${D}${datadir}/dbus
>> }
>>
>> Actually if the standard do_install needs to be changed, it generally
>> raises some alarms with me.
>> Why would we need to install something that the package normally does
>> not install?
>> (of course this could be a bug in the package, but then I would prefer
>> a patch that we also try to push upstream).
>>
>
> Agreed, that is odd.  If it's something that dbus using things might need,
> why is it only prepended for the native version, and as you say, why is it
> not installed by the upstream install target.  I'd do a git blame and figure
> out who added that and ask them :)

I guess it was part of converting dbus recipes to new staging in which case
it is converted but may not be used

> --
> Christopher Larson
> clarson at kergoth dot com
> Founder - BitBake, OpenEmbedded, OpenZaurus
> Maintainer - Tslib
> Senior Software Engineer, Mentor Graphics
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: virtclass_install prepend

[Frans Meulenbroeks]

2010/9/6 Chris Larson <clarson@kergoth.com>:
> On Mon, Sep 6, 2010 at 12:41 AM, Frans Meulenbroeks <
> fransmeulenbroeks@gmail.com> wrote:
>
>> 2010/9/6 Khem Raj <raj.khem@gmail.com>:
>> > On (19/08/10 21:14), Frans Meulenbroeks wrote:
>> >> 2010/8/19 Chris Larson <clarson@kergoth.com>:
>> >> > On Thu, Aug 19, 2010 at 4:29 AM, Frans Meulenbroeks <
>> >> > fransmeulenbroeks@gmail.com> wrote:
>> >> >
>> >> >> Hi,
>> >> >>
>> >> >> Found this in our reicpes:
>> >> >>
>> >> >> dbus/dbus-glib_0.86.bb:do_install_virtclass-native_prepend() {
>> >> >>
>> >> >
>> >> > This prepends to do_install_virtclass-native, which will then replace
>> >> > do_install when the overrides are applied.  This is prepending to a
>> variable
>> >> > which overrides, as opposed to the other, which is a conditional
>> prepend.
>> >> >  The two are quite different.  It's unlikely that this is what they
>> >> > intended.  Far more likely that it should be changed to
>> >> > _prepend_virtclass-native.
>> >>
>> >> I have no idea what is intended and I have no dbus knowledge, so can't
>> fix this.
>> >> I'll leave it to someone else to resolve this.
>> >
>> > I think the correct fix is as Chris suggested. Please go ahead and fix
>> it.
>> >
>>
>> Well I am a little bit hesitating as I am by no means an expert on the
>> dbus recipe.
>>
>> The other question that immediately popped up is:
>> If the code is wrong and no-one noticed, is it actually needed?
>>
>> The code is:
>> do_install_virtclass-native_prepend() {
>>         install -d ${D}${datadir}/dbus
>>        install -m 0644 dbus-bus-introspect.xml ${D}${datadir}/dbus
>> }
>>
>> Actually if the standard do_install needs to be changed, it generally
>> raises some alarms with me.
>> Why would we need to install something that the package normally does
>> not install?
>> (of course this could be a bug in the package, but then I would prefer
>> a patch that we also try to push upstream).
>>
>
> Agreed, that is odd.  If it's something that dbus using things might need,
> why is it only prepended for the native version, and as you say, why is it
> not installed by the upstream install target.  I'd do a git blame and figure
> out who added that and ask them :)
> --

Henning,
Git blame gives your name as the person who created those lines.
Care to comment?

Frans

Re: virtclass_install prepend

[Henning Heinold]

> >> The code is:
> >> do_install_virtclass-native_prepend() {
> >>         install -d ${D}${datadir}/dbus
> >>        install -m 0644 dbus-bus-introspect.xml ${D}${datadir}/dbus
> >> }
> >>
> >> Actually if the standard do_install needs to be changed, it generally
> >> raises some alarms with me.
> >> Why would we need to install something that the package normally does
> >> not install?
> >> (of course this could be a bug in the package, but then I would prefer
> >> a patch that we also try to push upstream).
> >>
> >
> > Agreed, that is odd.  If it's something that dbus using things might need,
> > why is it only prepended for the native version, and as you say, why is it
> > not installed by the upstream install target.  I'd do a git blame and figure
> > out who added that and ask them :)
> > --
> 
> Henning,
> Git blame gives your name as the person who created those lines.
> Care to comment?
> 
> Frans

Hi,

if I remember correctly this file is generated in the build process. But somehow
in the later versions of dbus-glib there is somehow an error so this file gets not installed
at all in the native variant. So I did it per hand.

Bye Henning

Re: virtclass_install prepend

[Richard Purdie]

On Mon, 2010-09-06 at 22:11 +0200, Henning Heinold wrote:
> > >> The code is:
> > >> do_install_virtclass-native_prepend() {
> > >>         install -d ${D}${datadir}/dbus
> > >>        install -m 0644 dbus-bus-introspect.xml ${D}${datadir}/dbus
> > >> }
> > >>
> > >> Actually if the standard do_install needs to be changed, it generally
> > >> raises some alarms with me.
> > >> Why would we need to install something that the package normally does
> > >> not install?
> > >> (of course this could be a bug in the package, but then I would prefer
> > >> a patch that we also try to push upstream).
> > >>
> > >
> > > Agreed, that is odd.  If it's something that dbus using things might need,
> > > why is it only prepended for the native version, and as you say, why is it
> > > not installed by the upstream install target.  I'd do a git blame and figure
> > > out who added that and ask them :)
> > > --
> > 
> > Henning,
> > Git blame gives your name as the person who created those lines.
> > Care to comment?
> > 
> > Frans
>
> if I remember correctly this file is generated in the build process. But somehow
> in the later versions of dbus-glib there is somehow an error so this file gets not installed
> at all in the native variant. So I did it per hand.

I suspect that code could well have come from Poky. In this case the
native recipe generates a file which is then used by the target builds.
It can't be generated in the target recipe as that would mean running
target binaries.

So it is in fact a perfectly valid thing to do in this case.

Cheers,

Richard

Re: virtclass_install prepend

[Khem Raj]

On Tue, Sep 7, 2010 at 6:57 AM, Richard Purdie <rpurdie@rpsys.net> wrote:
> On Mon, 2010-09-06 at 22:11 +0200, Henning Heinold wrote:
>> > >> The code is:
>> > >> do_install_virtclass-native_prepend() {
>> > >>         install -d ${D}${datadir}/dbus
>> > >>        install -m 0644 dbus-bus-introspect.xml ${D}${datadir}/dbus
>> > >> }
>> > >>
>> > >> Actually if the standard do_install needs to be changed, it generally
>> > >> raises some alarms with me.
>> > >> Why would we need to install something that the package normally does
>> > >> not install?
>> > >> (of course this could be a bug in the package, but then I would prefer
>> > >> a patch that we also try to push upstream).
>> > >>
>> > >
>> > > Agreed, that is odd.  If it's something that dbus using things might need,
>> > > why is it only prepended for the native version, and as you say, why is it
>> > > not installed by the upstream install target.  I'd do a git blame and figure
>> > > out who added that and ask them :)
>> > > --
>> >
>> > Henning,
>> > Git blame gives your name as the person who created those lines.
>> > Care to comment?
>> >
>> > Frans
>>
>> if I remember correctly this file is generated in the build process. But somehow
>> in the later versions of dbus-glib there is somehow an error so this file gets not installed
>> at all in the native variant. So I did it per hand.
>
> I suspect that code could well have come from Poky. In this case the
> native recipe generates a file which is then used by the target builds.
> It can't be generated in the target recipe as that would mean running
> target binaries.
>
> So it is in fact a perfectly valid thing to do in this case.

wow cool. This needs to be in the comments in that code.

>
> Cheers,
>
> Richard
>
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>