clang in jenkins

Alan Griffiths alan.griffiths at canonical.com
Tue Mar 26 09:24:34 UTC 2013


On 26/03/13 07:26, Daniel van Vugt wrote:
> You'd be hard-pressed to find a good argument against automated static
> analysis. It's always useful to improve code quality from more angles.
>
> I believe Alan's concern was that we were devoting resources to the
> effort that we couldn't spare... ?

I would have no objection to having clang in CI if all that was
happening is that it detected new errors. But there is more work needed
than that - and it is the timing of that work that concerns me. I think
after Daniel's work on clang support that the code we've written is
pretty close to being clean under clang.

However, the same is not true of the input stack we've forked from
android - and I don't think we should be trying to clean up that code
until we have it under automated tests. (FWIW There's other clean-up to
that code we've also postponed - adding clang to that exercise is
probably better timing.)

So, while long term I think clang support would be a good thing and I'd
keep it in the backlog, I don't think it something we should be spending
time on in the current iteration. (I'd rather we stabilised our arm
support.)



More information about the Mir-devel mailing list