Re: 'master' 'make check' lcov code coverage report

On Fri, Nov 7, 2014 at 7:29 PM, Norbert Thiebaud <nthiebaud@gmail.com>
wrote:
> On Fri, Nov 7, 2014 at 10:38 AM, Maarten Hoes <hoes.maarten@gmail.com>
wrote:
> > On Mon, Nov 3, 2014 at 8:12 PM, Maarten Hoes <hoes.maarten@gmail.com>
wrote:
> > > On Mon, Nov 3, 2014 at 8:02 PM, Markus Mohrhard
> > > <markus.mohrhard@googlemail.com> wrote:
> > > > Maybe not weekly but bi-weekly or monthly. Otherwise it will only be
> > > > done from time to time without a plan.
> > > Hrm. Good point.
> > >
> > > > Even weekly would not hurt as
> > > > we have neither a problem with disk space or build hardware.
> > > Cool.
> > >
> > > > You can choose whatever technology is a good fit and that you know.
Of
> > > > course something that other people can understand and maintain is
> > > > preferred. A shell script sounds like a good idea for this task and
> > > > would simplify the handling.
> > > Alright, I guess that means Ill have a go at it, then. So who do I
need to
> > > mail my pub ssh key and what is the hostname/ip of the machine I would
be
> > > allowed to royally mess up for this purpose ?
> > > ;)
> >
> > Hi,
> >
> >
> > So I have something now that seems to work (doing a first test run now)
on
> > the vm cloph kindly provided, but has no logging, error notification,
source
> > updating, error handling, or uploading of reports.
> >
> >
> > Time for a few questions:
> >
> >
> > 1.)
> > Where do the reports have to be uploaded to ?
> >
> > 2.)
> > What do you want the script to do when it detects a failure ? Log an
error ?
> > Send an email to an admin ?
> >
> > 3.)
> > Does the script have to handle its own source code updating ? git pull ?
>
> all these 3 point could be solved by putting that job under jenkins
> control.. I'll have a look at doing that.
Alright, so no complex error handling code then. But I guess you still will
want the script to exit with a non-zero exit code if a single one of the
commands in it fails (so jenkins can pick up on the failure) instead of it
just blindly continuing as if nothing happened ?


> > 4.)
> > If the make check (run the tests) fails, do you want the script to:
>
> Ideally I would chain that job behind a sucessful build, so that this
> does not happen
> iow first fidn a 'good' commit, then submit the job....
> but before that happen, the msot usefull way would be I think to just
> drop that build, and retry a bit latter with hopefully a good build at
> that time
So in other words: instead of 'make -k check' do a 'make check', and have
the script exit with a non zero exit code if the make command fails ?


> > 6.)
> > Should I send a license/code statement to the list for the 'code', even
> > though this is just a shell script ?
>
> This is not really part of the product we package... os just put an
> appropriate open-source license statement at the to of the script.
> and prolly the best is to find a place for it in 'buildbot.git'
> (accessble via gerrit)
So where does this 'buildbot.git' live, and how do I set up submit access
to that so I can commit the script ?



- Maarten