KDE/CI for KF5 WAS "OSX/CI: libkcddb and ark fail to build for KF5"

[Ben Cooksley]

On Sun, Mar 8, 2015 at 12:43 AM, Marko Käning <mk-lists at email.de> wrote:
> Hi Jeremy AND folks,

Hi Marko,

>
> On 07 Mar 2015, at 02:14 , Jeremy Whiting <jpwhiting at kde.org> wrote:
>> I appreciate the efforts you are making :)
>
>
> Thanks. :)
>
> Just wondering whether I am wasting my time because of my “perfectionism" here on OSX/CI.

Not entirely. What we probably need is a better way for developers to
communicate to people on the porting status.
The best way to do this would probably be through the build metadata -
if the branch has been set then we can probably assume developers
think it is ready.

> E.g. libkcddb I perhaps shouldn’t have touched at all for now, as it hasn’t been committed to
> since last autumn… Well, for ark it’s different - of course!
>
> But still, e.g. on my dependencies-RR [1] I have been working for 2 weeks now and while doing
> so I more and more began to wonder whether it makes sense for all the new projects I am trying
> to get into CI...

Anything being released needs to have CI really. If the tarballs end
up on download.kde.org, it should be Green on CI first.

>
> I have the feeling that my work (inspired by Christoph Feck’s post on KDE-DEVEL [2] and his
> porting status page [3]) runs a little too uncoordinated - which gives me only extra head aches,
> since I have to iterate through all these projects _manually_ while knowing far too little of their
> real status and objectives. This lead to the result that I
>
>         - could effectively only add about _30_ new projects
>
>         - which means that there are now about 200 projects successfully built on OSX/CI [4],
>
>         - while having had to go through *170* potential projects from [3] step by step.
>
> For all of these I needed to figure out how to build them on OSX/CI and how to fix their
> dependency metadata. 140 projects are thus disabled on OSX/CI for now. I doubt that this can
> be considered a very efficient workflow! ;-)

That is quite a bit of effort for little return.
Out of interest, what was the primary blocker to getting things to
build? 140 seems quite high as a number....

>
> I think we need to have another CI IRC meeting sometime soon, where we could discuss -
> besides the change to the new KDE/CI system - also which projects are actually worth
> considering (not only on OSX but) on Linux/CI in general!

Anything in Extragear, Frameworks or kde/* should be covered by CI
really. If it can't build, we need to fix that.

>
> What do you think?
>
> Greets,
> Marko
>
>
>
> [1] https://git.reviewboard.kde.org/r/122672/
> [2] http://lists.kde.org/?l=kde-devel&m=142434697530863&w=2
> [3] http://developer.kde.org/~cfeck/portingstatus.html
> [4] https://trac.macports.org/wiki/KDEProblems/KDEMacPortsCI/Status

Cheers,
Ben