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

Bug#798955: Moving glibc headers from /usr/include to /usr/include/$(DEB_HOST_MULTIARCH)



On 2015-09-23 21:38, Aurelien Jarno wrote:
> On 2015-09-14 15:13, Aurélien Gérôme wrote:
> > Source: glibc
> > Version: 2.21-0experimental1
> > Severity: wishlist
> > Tags: patch moreinfo
> > 
> > Hi,
> > 
> > On a multiarch-enabled Debian system, there is an impossibility to
> > install libc0.1-dev, libc0.3-dev, libc6-dev, and libc6.1-dev binary
> > packages at the same time, because glibc headers are all installed
> > in /usr/include.  Therefore, it does not allow us to install (in a
> > simple multiarch manner) several dependencies needed to cross-compile
> > for another target, e.g. cross-compiling for an amd64 target on a
> > hurd-i386 host.
> > 
> > This patch moves those headers from /usr/include to
> > /usr/include/$(DEB_HOST_MULTIARCH) and removes conflicts between
> > libc0.1-dev, libc0.3-dev, libc6-dev, and libc6.1-dev binary packages.
> 
> That looks fine on the principle, though that unfortunately prevent 
> sharing files between packages. I guess at some point we'll want to
> provide an arch:all libc-dev-common file to put them. That said we
> have to do the change anyway, as musl people will request it.
> 
> > So far, Helmut rebuilt ~100 packages fine with that include-moved
> > glibc.  Right now, we are waiting for a Debian access to Amazon
> > instances to rebuild the entire archive against that patched glibc
> > and to deal with possible FTBFS.  Tracking can be done on [0].
> 
> Before doing such a change, I clearly want to make sure it won't
> introduce (unknown) regressions. Doing an archive rebuild is clearly the
> way to go, thanks for doing that.

A small update from IRC:

23:52 < helmut> aurel32: quick update on #798955: 1) we know that it
breaks any package that uses multilib 2) we know that it breaks gcc's
fixincludes 3) we know that it breaks python2.7
23:53 < helmut> aurel32: i.e. sorting it out will take longer

-- 
Aurelien Jarno                          GPG: 4096R/1DDD8C9B
aurelien@aurel32.net                 http://www.aurel32.net


Reply to: