[bind10-dev] failed unittests
JINMEI Tatuya / 神明達哉
jinmei at isc.org
Thu May 24 18:37:09 UTC 2012
At Wed, 23 May 2012 09:15:19 +0200,
Jelte Jansen <jelte at isc.org> wrote:
> > So, the patch pasted below solves the problem on this particular
> > platform. I'm not sure if this is a generally applicable solution
> > or can be safely applied without breaking others, though.
>
> I think it can be pretty safely applied, and I think it would be a
> pretty easy task to scan through all makefiles and apply this
> everywhere. IIRC we had a few ordering problems before, but those were
> not internal-libs-vs-dependency-ordering.
>
> However, it boils down to the question whether we want to start using
> library versioning for all libs now, or not. If we do, I actually
> propose we do the exact opposite, and let this be a canary for version
> increases (don't upgrade lib version until this tester fails). It
> wouldn't catch all possible problems, but it would for most of them.
>
> If we do not want to start using versioned libraries now, the below
> patch is indeed what I think we should do.
As you probably noticed I've applied it (with one minor cleanup), and
also addressed a similar issue in a different place that way, as it
broke buildbots and I thought it was urgent.
I don't yet understand how exactly the versioning works in this
context, but if we do versioning and it solve this problem also, I
have no problem with reordering the libs again so that they can work
as a canary.
---
JINMEI, Tatuya
More information about the bind10-dev
mailing list