RE: [late-review] Re: libar Makefile.am,1.9,1.10 ar.3,1.1,1.2 ar.c,1.7,1.8 ar.h,1.2,1.3

From: Murray S. Kucherawy <msk_at_cloudmark.com>
Date: Wed, 1 Sep 2010 14:35:49 -0700

> -----Original Message-----
> From: opendkim-dev-bounce_at_lists.opendkim.org [mailto:opendkim-dev-
> bounce_at_lists.opendkim.org] On Behalf Of Mike Markley
> Sent: Wednesday, September 01, 2010 12:38 PM
> To: opendkim-dev_at_lists.opendkim.org
> Subject: Re: [late-review] Re: libar Makefile.am,1.9,1.10 ar.3,1.1,1.2
> ar.c,1.7,1.8 ar.h,1.2,1.3
>
> In my mind, the key issue for whether a new ABI version is needed is
> whether applications linked against the old one will break with the new
> one. So if doing it the old way will cause an issue, or if the function
> signature has changed, or if the output behavior is otherwise different
> now than it was before, then it definitely makes sense. On the other
> hand, if a program linked against libopendkim.so.2 will continue to
> work
> if this new library is laid down in its place, then it's probably not
> necessary.
>
> It's also not hugely onerous to bump the version number, of course, but
> that means a recompilation will be needed to take advantage of future
> bugfixes, whether those change the ABI or not.

I'm more conservative about bumping the version of libopendkim, but in this case we're talking about libar which is optional to the product anyway.

That said, if people still think we shouldn't increase the libar.so version given the described change, I'll back it down.
Received on Wed Sep 01 2010 - 23:08:42 PST

This archive was generated by hypermail 2.3.0 : Mon Oct 29 2012 - 23:32:54 PST