[OE-core] [PATCH] combo-layer: make Signed-off-by optional
Paul Eggleton
paul.eggleton at linux.intel.com
Fri Mar 13 08:57:35 UTC 2015
On Thursday 12 March 2015 20:45:32 Patrick Ohly wrote:
> On Thu, 2015-03-12 at 18:21 +0000, Paul Eggleton wrote:
> > On Monday 09 March 2015 13:56:39 Patrick Ohly wrote:
> > > +# global options
> > > +[DEFAULT]
> > > +
> > > +# Add 'Signed-off-by' to all commits that get imported automatically.
> > > +signoff = True
> > > +
> > >
> > > # component name
> > > [bitbake]
> > > # mandatory options
> >
> > So I'm OK with adding this in as an option. However to me a name like
> > DEFAULT implies you're establishing a general section to apply default
> > settings for all components where the component can override those
> > defaults if it chooses, which doesn't really represent what this does -
> > so a different name might be more appropriate (GLOBAL or _global_
> > perhaps?)
>
> "DEFAULT" is the special string that Python's ConfigParser uses, well,
> by default for that special section which does not show up in the list
> of sections. I don't know how to rename that.
>
> I'm probably abusing this concept a bit here: it seems that special
> section is meant to provide default values that get returned also for
> the other sections when they don't have their own value.
>
> Here's a cleaner solution:
> * Get rid of the [DEFAULT] section in the file.
> * When reading it, on-the-fly prepend the string
> '[this-is-not-really-a-repo'].
> * When reading global properties, get it from that section.
> * When listing sections to find repos, ignore it.
>
> How about that?
Hmm, ok, I wasn't aware of this built-in "DEFAULT" section behaviour. A simple
solution then would just be to make this setting per-component and keep the
DEFAULT usage in the example config - although I do realise it's unlikely to be
very useful on a per-component basis, at least it will work in the same way as
other values we read from there.
Cheers,
Paul
--
Paul Eggleton
Intel Open Source Technology Centre
More information about the Openembedded-core
mailing list