On 2010-04-17T21:04:06, Lars Ellenberg <[email protected]> wrote:
> This is the interdiff:
>
> Bump the so version, note this is the libtool
> "interface:revision:age" way, the resulting soname
> is libplumb.so.2.1.0
>
> --- a/lib/clplumbing/Makefile.am Thu Apr 15 15:58:50 2010 +0200
> +++ b/lib/clplumbing/Makefile.am Sat Apr 17 20:31:18 2010 +0200
> @@ -63,7 +63,7 @@
>
> libplumb_la_LIBADD = $(top_builddir)/replace/libreplace.la \
> $(top_builddir)/lib/pils/libpils.la
> -libplumb_la_LDFLAGS = -version-info 2:0:0
> +libplumb_la_LDFLAGS = -version-info 3:0:1
I must be completely silly, but from 2:0:0 to 3:0:1 doesn't seem to
imply anything about 2.1.0?
> Introducing the new API in an ABI incompatible way without
> reflecting that in the soname was a bug.
> We all agree about that.
Yes. The segfaults in 1.0.4 clearly have been inacceptable.
> Meanwhile I had time to review and test various approaches,
> and find that the first three chunks posted above make for
> an equivalent API without breaking ABI backwards compatibility.
The above will force users to upgrade, at least if they are using sane
package managers; rpm will notice that 2.x.x is no longer provided.
> What do we disagree about, again?
I disagreed about anything that needed a runtime test for code that
wanted to use this, and compatibility wrappers like you propose.
So the upgrade of libglue2 requires a rebuild; what's the problem with
that? The cluster is already down ANYWAY, a point which you completely
ignore and STILL haven't replied to.
Regards,
Lars
--
Architect Storage/HA, OPS Engineering, Novell, Inc.
SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
"Experience is the name everyone gives to their mistakes." -- Oscar Wilde
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
