Am Sonntag, den 26.06.2011, 13:25 -0300 schrieb Evandro Giovanini:
On Sun, Jun 26, 2011 at 1:01 PM, Christoph Wickert
<christoph.wickert(a)googlemail.com> wrote:
> Am Sonntag, den 26.06.2011, 17:08 +0200 schrieb Kevin Kofler:
>> Felix Miata wrote:
>> > FF5 is the security update to FF4.0.1, which incorporates an upstream
>> > versioning policy change.
>>
>> The funny thing is that Firefox is going exactly the opposite way of us with
>> their update policies,
>
> They didn't change their update policy but their release/development
> model. FF 5 is an update to FF 4, but 3.6 got an update to 3.6.18, too.
> This means that Mozilla's update policy hasn't changed.
>
>> and that as a result, that Firefox security update is
>> not compliant with our update policies
>
> At what point exactly? Basically all that has changed is the version,
> IHMO FF 5 can better be described as FF 4.1. The user experience hasn't
> changed and the update meets all requirements of the update policy, so I
> really don't see a problem here.
>
Firefox 5 is not stable because it introduces new features.
New features don't have anything to do with stability. It's the AI/ABI
and AFAIK nothing has changed in xulrunnner. All dependent packages
required a simple rebuild. We had these rebuilds with every single FF
update.
The most
visible example of this is users who had extensions stop working with
Firefox 5.
This is most likely because the extensions expect a certain version
string but not because of API/ABI changes.
Firefox's policy has changed. In the past they supported a
stable
version for more than one day after the new release was out.
What policy? The security policy? Firefox 3.6 is considered the previous
stable version and it is still supported.
They're
not doing that anymore with Firefox 4, so users are forced to use the
new features (and bugs) of a new release. It's a great policy if
you're in a race to not lose mindshare from Chrome, it's not so great
for the people who have Firefox deployed in stable environments.
Look, I never said I like the new version scheme. In fact I said that I
dislike it. The major version shouldn't have been bumped because is a
stable update, this means there are no API/ABI changes and no change in
configuration formats either. There are merely no changes and there is a
clean upgrade path, so I have no idea what problems in a stable
environment you are expecting.
And I have no idea what part of our update policy should be violated by
this update. Please somebody enlighten me.
Regards,
Christoph