Re: pushing patches despite requests for a test

He Kendy,

On Fri, Jul 24, 2015 at 9:14 PM, Jan Holesovsky <kendy@collabora.com> wrote:

> Hi Bjoern, Moggi,
>
> Bjoern Michaelsen píše v Pá 24. 07. 2015 v 19:51 +0200:
>
> > > On Friday, 2015-07-24 18:22:34 +0200, Markus Mohrhard wrote:
> > > > I'd appreciate a decision by the ESC to either stop this behavior or
> to
> > > > tell me to shut up! In the second case I will take it as another
> sign that
> > > > the project does not care about quality anymore.
>
> Sorry that you feel offended by this; but please let's not overreact :-)


I mentioned the last time that this happened in a similar fashion to
Michael that I think that it is a really bad move. Now it happened again
and again more or less in the same way.

Additionally I consider the "it's insane to rejecting this on the grounds
of unit tests" a quite direct personal attack. Normally I would have tried
to talk directly with someone who pushes such a patch but after such a
statement which is clearly targeted at me I prefer that the conversation
happens in the open.

I stand by my decision not to accept any chart patches that have no unit
test (currently pure dialog code is excluded as there is no sane way to
test it). If I don't manage to review a patch and someone else pushes it it
is tough luck for me but if I find the time and energy to review a patch
and request a test I consider it extremely rude and inappropriate to push
it nevertheless.


> > > Of course we do care about quality, and everyone should. Education
> > > sometimes needs time and annoying repetition..
> >
> > Yes. Lets reiterate this on the ESC. Personally, I am quite in support
> of a
> > "think how to get more tests and more coverage" stance. This is indeed a
> > mindset question first.
>
> In this particular case, Tomaž gave explanation why he decided to push
> the patch even without the unit test, and also promised that he'll come
> up with a way how to test it reliably:
>
> "First I want this patch in.. later I'll try to find a way to reliably
> test this."


> I recall that Tomaž bootstrapped test infrastructure in several modules
> & areas & new ways, so I trust him that he will keep the promise ;-) -
> so I don't really see a problem with this approach here...


The correct thing would have been to talk to me. I would have helped with
the XShape tests as I did for anyone else who has needed some help with
tests.


> We can go through that at the ESC of course; but until then - we should
> at least get a friendly message to Laurent who's the author of the
> original patch, and a poor victim here :-)  I'll try to ressurect the
> gerrit issue entry in the meantime so that Laurent knows.


I'd appreciate if it is discussed in the ESC as I think that no patch
should be pushed if a core developer likes to see a test or some more
discussion.

Regards,
Markus