[oe] Examples of srctree and gitver
Ambrose, Martin
martin at ti.com
Thu Jun 3 13:55:08 UTC 2010
On Wed, Jun 02, 2010 at 16:47:17, Ambrose, Martin wrote:
> On Tue, Jun 01, 2010 at 08:36:40, Michael Smith wrote:
>
> > Ambrose, Martin wrote:
> > > On Sun, May 30, 2010 at 14:49:32, Chris Larson wrote:
> > >> Aside: the way it does so is quite the hack, we need to figure out
> > >> a better way to manipulate the task graph like this.
> > >
> > > Maybe that explains the problem I'm seeing with the dependency graph when using srctree.
> >
> > > Then it goes straight to the configure phase without ensuring the dependencies are present.
> >
> > Yes, I see similar things. I'm still on an ancient tree with
> > do_populate_staging(). If memory serves, the prerequisite packages are
> > unpacked, compiled and installed, but do_stage isn't run before the
> > srctree recipe's do_configure.
> >
> > As long as you're not doing parallel builds it's easy enough to work
> > around: just put the libraries you need in your IMAGE_INSTALL before the
> > srctree packages.
>
> Thanks for the pointer. Working around for now by building the dependencies
> by hand so I could get to the next problem. This being that my do_install function
> is not being populated from what's in my recipe.
>
> My recipe has
>
> do_install() {
> oenote "RUNNING install: ${D}"
> install -d ${D}/home/root/appdir
> install -m 0755 ${S}/myapp ${D}/home/root/myapp
> }
Actually my closing bracket here "}" wasn't in column zero.
My emacs python mode doesn't do proper indentation I guess.
Anyway this seems to be a python/parsing error, with a silent failure,
and nothing to do with srctree AFAICT.
Regards,
Martin
>
> But the resultant run.do_install script generates an empty function
>
> base_do_install() {
> :
>
> }
> ...
> do_install() {
> base_do_install
>
> }
> ...
> cd /home/user/mysrc
> do_install
>
>
> Looks like the .bbclass which generates the script is not happy about something
> but not sure where to look to find out what I'm missing.
>
> Regards,
> Martin
>
Regards,
Martin
More information about the Openembedded-devel
mailing list