This is the mail archive of the
newlib@sourceware.org
mailing list for the newlib project.
Re: Bug#338059: acknowledged by developer (Re: Bug#338059: newlib_1.13.0-2(armeb/unstable): FTBFS: please add armeb to debian/control)
- From: Shaun Jackman <sjackman at gmail dot com>
- To: newlib at sources dot redhat dot com
- Cc: Wouter Verhelst <wouter at debian dot org>, 338059-forwarded at bugs dot debian dot org, Debian Bug Tracking System <control at bugs dot debian dot org>
- Date: Tue, 8 Nov 2005 09:32:59 -0700
- Subject: Re: Bug#338059: acknowledged by developer (Re: Bug#338059: newlib_1.13.0-2(armeb/unstable): FTBFS: please add armeb to debian/control)
- References: <7f45d9390511071658o22af5d3bv@mail.gmail.com> <20051108001237.GH15884@country.grep.be> <handler.338059.D338059.11314115026058.notifdone@bugs.debian.org> <20051108012514.GA2431@country.grep.be>
- Reply-to: Shaun Jackman <sjackman at gmail dot com>
package newlib
tag 338059 +confirmed upstream
thanks
Hello Wouter,
I'll forward your request on to the newlib folks. My reply follows the
context quoting. Please follow up to newlib@sources.redhat.com,
338059@bugs.debian.org.
2005/11/7, Wouter Verhelst <wouter@debian.org>:
> On Mon, Nov 07, 2005, Shaun Jackman wrote...
> > i386 is the only Linux target supported by the upstream newlib. Many
> > other embedded targets are supported by newlib, but not *-linux, and
> > in particular not armeb-linux. This is not likely to change anytime
> > soon, though it would possible to write a port for other Linux targets
> > if someone wished to contribute the work.
>
> In that case, wouldn't it be better to make the build system bail out
> when it detects it's on an architecture it doesn't support? (rather than
> allowing to build everything and only at the very end finding out that
> hey, this architecture isn't actually supported).
>
> (Reopening as I think that'd be better, but this'll be the last time.)
The issue is that newlib is one component of the GNU toolchain build
system. Even though newlib fails to build on armeb-linux, other
components such as libiberty may succeed.
The technical reason aside, however, I agree with you that it would be
better behaved if the "newlib" tarball distribution failed to build if
newlib fails to configure.
Cheers,
Shaun