Re: [PATCH 2/2 v2] archive-*-source.bbclass: Handle all package classes

[Richard Purdie <richard.purdie@linuxfoundation.org>]

On Mon, 2013-01-14 at 18:48 +0500, Shakeel, Muhammad wrote:
> From: Muhammad Shakeel <muhammad_shakeel@mentor.com>
> 
> * Add archiver tasks correctly for any 'IMAGE_PKGTYPE'.
> 
> [YOCTO #3449]
> 
> Signed-off-by: Muhammad Shakeel <muhammad_shakeel@mentor.com>
> Signed-off-by: Noor Ahsan <noor_ahsan@mentor.com>
> Signed-off-by: Christopher Larson <chris_larson@mentor.com>
> ---
>  meta/classes/archive-configured-source.bbclass |   15 +++++++++++++--
>  meta/classes/archive-original-source.bbclass   |   15 +++++++++++++--
>  meta/classes/archive-patched-source.bbclass    |   15 +++++++++++++--
>  3 files changed, 39 insertions(+), 6 deletions(-)
> 
> diff --git a/meta/classes/archive-configured-source.bbclass b/meta/classes/archive-configured-source.bbclass
> index 64799e3..50838e6 100644
> --- a/meta/classes/archive-configured-source.bbclass
> +++ b/meta/classes/archive-configured-source.bbclass
> @@ -11,12 +11,19 @@ inherit archiver
>  addtask do_archive_configured_sources after do_configure
>  
>  # Get archiving package with temp(logs) and scripts(.bb and inc files)
> -addtask do_archive_scripts_logs after do_package_write_rpm
> +addtask do_archive_scripts_logs
>  
>  # Get dump date and create diff file 
> -addtask do_dumpdata_create_diff_gz after do_package_write_rpm before do_build
> +addtask do_dumpdata_create_diff_gz before do_build
>  
>  python () {
> +    packaging = d.getVar('IMAGE_PKGTYPE', True)
> +    write_package_task = 'do_package_write_' + packaging
> +
> +    deps = d.getVarFlag('do_dumpdata_create_diff_gz', 'deps') or []
> +    deps.append(write_package_task)
> +    d.setVarFlag('do_dumpdata_create_diff_gz', 'deps', deps)

I'd not previously noticed bitbake was exposing this "deps" flag as a
list. I'd previously assumed we interacted with strings through the
deptask flag so we'd use something like:

d.appendVarFlag('do_dumpdata_create_diff_gz', 'deptask', ' do_package_write_' + packaging)

We should probably make a decision whether deps is meant to be exposed
or whether we should be using deptask. I appreciate this patch isn't at
fault, its the archiver code in general doing this. I agree the API is
less optimal using strings but it is also more consistent with every
other set of flags...

Cheers,

Richard