On Mon, Sep 07, 2015 at 01:50:35PM +0200, Alexander Kurtz wrote:
> Hi,
>
> On Mon, 2015-09-07 at 13:00 +0200, Andreas Henriksson wrote:
> > I think we won't touch the soname, since we usually try to not
> > deviate from what upstream uses but only change the package
> > name and add conflicts against old package name...
> > (This solves packaged software upgrades, but leaves self-compiled
> > binaries out in the cold.)
> > Same as was done last time the ABI broke in similar way.
>
> This seems very reasonable. Do you agree with my assessment, that
> libical *is* actually ABI-stable now and should not break compatibility
> in future versions, even if new entries are added to the various enums?
I've not looked to closely at what will happen in future versions.
I poked Simon McVittie on IRC trying to sucker him into this discussion
as he has been involved in the past and also handled similar issues
in dbus-glib... He summarized as:
13:20 < smcv> I don't think there's any solution that doesn't involve
committing a list of the enum members in their canonical order to
the VCS
If he says so, then I'm doubting libical has found one....
Regards,
Andreas Henriksson
