Re: [PATCH] buildstats-summary.bbclass: Import useful bbclass from meta-mentor

[Martin Jansa <martin.jansa@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1728 bytes --]

On Wed, Jun 11, 2014 at 01:54:38PM -0700, Christopher Larson wrote:
> On Fri, Jun 6, 2014 at 12:27 PM, Martin Jansa <martin.jansa@gmail.com>
> wrote:
> 
> > Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
> > ---
> >  meta/classes/buildstats-summary.bbclass | 43
> > +++++++++++++++++++++++++++++++++
> >  1 file changed, 43 insertions(+)
> >  create mode 100644 meta/classes/buildstats-summary.bbclass
> >
> > diff --git a/meta/classes/buildstats-summary.bbclass b/meta/classes/
> > buildstats-summary.bbclass
> > new file mode 100644
> > index 0000000..3c02724
> > --- /dev/null
> > +++ b/meta/classes/buildstats-summary.bbclass
> > @@ -0,0 +1,43 @@
> > +# Imported from meta-mentor
> > +# http://git.yoctoproject.org/cgit/cgit.cgi/meta-mentor/log/classes/
> > buildstats-summary.bbclass
> > +# with following modifications
> > +# 1) when both setscene and normal tasks were executed, count it as
> > no_sstate
> > +# 2) count do_populate_sysroot and do_packagedata tasks, because often we
> > invalidate only runtime deps
> >
> 
> Looks like useful improvements, nicely done. Have you thought about not
> hardcoding the task list, if we're going to display it this way? See
> https://gist.github.com/kergoth/2519927f59712282e9e8 for a version which
> obeys SSTATETASKS.

I'm fine with your version showing all SSTATETASKS, would you mind
sending PATCHv2 with your modification or should I?

I've hardcoded packagedata only because it was typical case when I was
wondering why the build took so long even with populate_sysroot reused
from sstate (e.g. after some package.bbclass change), but showing all
doesn't hurt.

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 188 bytes --]