[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#1010755: qbs: FTBFS on riscv64


On Mon, 9 May 2022 at 11:51, John Paul Adrian Glaubitz
<glaubitz@physik.fu-berlin.de> wrote:
> Hi Bo!
> On 5/9/22 16:27, Bo YU wrote:
> s error just means that the symbols file(s) have to be updated, see [1].
> >>
> > Before I filed the bug, I realized the ftbfs issue might be with the
> > symbols file as you point.
> > `apt source qbs` and the debian/symbols file has the symbols:
> > ```
> > _ZN6QDebuglsEPKc@Base 1.22.0
> >  _ZN6QDebuglsERK7QString@Base 1.22.0
> >  _ZN7QStringD1Ev@Base 1.22.0
> >  _ZN7QStringD2Ev@Base 1.22.0
> > ```
> > It seems the symbols file is ok and I try to build it on riscv64
> > hardware locally and it is ok :)
> >
> > Or I am misunderstanding the buildd log error hint?
> It might be that the symbols file mismatch is considered a fatal error on the buildd but
> not for your local builds. Or the buildd was building with different build dependencies,
> we can try to trigger a rebuild on the buildd.

Missing non-optional symbols is fatal. But in this case:

$ c++filt _ZN6QDebuglsERK7QString
QDebug::operator<<(QString const&)


$ grep -Rn QDebug /usr/include/qbs/*
/usr/include/qbs/tools/codelocation.h:90:QDebug operator<<(QDebug
debug, const CodeLocation &location);

As you can see the signatures do not match, so the missing symbol must
come from some private header and it's not exposed to the user.

The maintainer will probably add the required changes in the next
upload, as usual.

Lisandro Damián Nicanor Pérez Meyer

Reply to: