This is the mail archive of the
mailing list for the elfutils project.
Re: [COMMITTED] libebl.h: Add comment from README that this is completely UNSUPPORTED.
- From: Mark Wielaard <mjw at redhat dot com>
- To: elfutils-devel at lists dot fedorahosted dot org
- Date: Tue, 27 Jan 2015 16:10:56 +0100
- Subject: Re: [COMMITTED] libebl.h: Add comment from README that this is completely UNSUPPORTED.
On Tue, Jan 27, 2015 at 08:46:11AM -0600, David Smith wrote:
> > + The ABI of the backend modules is not guaranteed. Really, not guarantee
> > + whatsoever. We are enforcing this in the code. The modules and their
> > + users must match. No third-party EBL module are supported or allowed.
> > + The only reason there are separate modules is to not have the code for
> > + all architectures in all the binaries. */
> The comment is fine (although I think you meant "Really, no guarantee"
> instead of "Really, not guarantee").
Thanks. I silently fixed this in both the header and the README,
> However, what about making this
> internal header more internal? Does code using elfutils need anything
> out of this file? If not, why does it end up in /usr/include/elfutils?
Some distributions (e.g. Fedora) package elfutils libasm. Unfortunately
libasm isn't really ready to be declared stable. In particular asm_begin ()
takes an Ebl *handle to initialize the AsmCtx_t *handle with. There is no
way to get an Ebl *handle except by using the libebl.h header.
As was recently discussed we might just have
to break this and update the .so version to show we made a mistake.
Ulrich, do you have any more comments on that discussion? Are there any
other changes desired if we are going to break ABI anyway?