Re: Review, please ! e: [med-svn] [rapmap] branch upstream updated (5fb69ea -> b8057c0)
On 23.10.17 16:46, Andreas Tille wrote:
> On Mon, Oct 23, 2017 at 04:17:37PM +0200, Steffen Möller wrote:
>> I admit my motivation was mostly to make yet another "New Version
>> Available" button on our tasks page go away ... and add another link to
>> the registries.
> That's a perfectly valid motivation and pushes me from time to time to
> work down this list of yellow buttons (if we do not have some R
> migration or lots of open bugs). Everybody should just pick from our
> tasks page some interesting target. In most cases such upgrades are
> really straigthforward.
Args. Well. Kind of. The problematic bits are in debian/patches.
I admit to have been bad at this myself, but even in the best of all
possible cases that upstream has adopted a patch, every patch to go
through, understand and adapt or decide to ignore - that is work. And it
happens at the wrong place. It should be on upstream's github.
>> Of course I agree with the test scripts. I truly believe though that we
>> need to think more about how to perform this testing on a workflow level.
> +1
Thanks. I am a bit stuck on how to treat data-only packages. We have
several ways on how to get public data maintained. But for testing, it
all needs to be a package already. So we may have to somehow intertwine
the data collection with the package formation with that data in place.
I am a bit shying off from any such automated package generation. This
may be something for volatile, though.
Best,
Steffen
Reply to: