Re: [gtk-vnc-devel] Releasing proposal



Jonh Wendell wrote:
See, users have complained about the focus issue, which was fixed after
0.3.3. So, the fix will be available in 0.3.4.

Then I can't just say: upgrade gtk-vnc to 0.3.4, because this new
version is too different from previous, with new symbols, and worse: new
behavior.

New symbols shouldn't be a problem. If your application is not using a symbol there's no issue.

I don't intend on introducing new behavior--but the behavior of set_pointer_grab() changed from 0.3.2 to 0.3.3 and it broke at least gvncviewer and kvm-test. We should have caught the change in behavior before the 0.3.3 release but it slipped through.

So for 0.3.4, I've simply reverted to the 0.3.2 behavior. I understand what you were trying to achieve when you changed the behavior of set_pointer_grab() but I think the better way to do that is with a new symbol which I've added for 0.3.4.

I'm certainly open to suggestion, the overriding goal is to break as few users as possible. We broke some users with the change from 0.3.2 to 0.3.3. If reverting to back to the 0.3.2 behavior breaks even more users, then we can do it as new symbols and preserve the 0.3.3 behavior.

So, I could guarantee to vinagre's users that it will work with gtk-vnc
0.3.x.
I'd like to keep users working as much as possible. If that means introduce API warts, I'm okay with that. Eventually, once we've got enough confidence that we know how to do it right, we'll break the API only once and get rid of the warts.

We were releasing 0.1, 0.2, 0.3.
0.3.1 came because 0.3.0 tarball was broken, and 0.3.1 was a fix only
release.

Why not continue with this approach?

I don't see why it matters much how the numbering goes. I don't want to get into the habit of frequently breaking the ABI though. Starting with 0.3.0, I think we're making a commitment not to break the ABI.

Regards,

Anthony Liguori

Cheers,





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]