[asterisk-users] 1.6.2.14 > 1.6.2.15: blind transfer works but not Xfer on aastra

[Kevin P. Fleming]

On 12/10/2010 04:18 PM, sean darcy wrote:
> On 12/10/2010 05:01 PM, Kevin P. Fleming wrote:
>> On 12/10/2010 03:26 PM, sean darcy wrote:
>>> On 12/10/2010 02:57 PM, Kevin P. Fleming wrote:
>>>> On 12/10/2010 01:45 PM, sean darcy wrote:
>>>>
>>>>> This was supposedly fixed in 1.6.2 on November 22, 2010. So isn't the
>>>>> fix in 1.6.2.15, released 12/8?
>>>>>
>>>>> In any event, that bug has been declared fixed, so you can't add a note.
>>>>
>>>> Not necessarily, no. Releases go through a 'release candidate' phase for
>>>> a week (or two, sometimes three) before being declared 'ready', so fixes
>>>> made before the release date aren't necessarily included. The changelog
>>>> included in the release will always indicate what revisions are included
>>>> in it, though.
>>>>
>>> 1.6.2.15-rc1 was released, or at least announced, on November 23. In any
>>> event, it'd seem that the purpose of rc's should be to catch regressions
>>> like this one.
>>
>> That is indeed the purpose; was the issue reported prior to 1.6.2.15
>> graduating to a full release? If not, that means nobody saw it, which is
>> unfortunate, but given that it's not realistic to expect hundreds of
>> users to test release candidates in real-world scenarios, it's what happens.
>>
>> This is also why the Asterisk test suite continues to grow, in order to
>> be able to catch regressions of this type before they even get into a
>> release candidate. If there's not an existing test that could catch this
>> problem, then that's an area where some help would be quite welcome.
>>
>
> Well, just to beat this dead horse more than it deserves, the point is
> that the regression in 1.6.2 was known, and fixed, on November 22. In
> other words, the day before rc1 was even announced.
> https://bugs.digium.com/view.php?id=18185#129038

But the 1.6.2.15-rc1 tag was made on 2010-11-15, one week earlier. 
Granted, a one week delay between the tag being made and being announced 
is a bit excessive, but it still completely explains why the fix was not 
in -rc1.

The 1.6.2.15 release was made on 2010-12-02, which certainly indicates 
that not being aware of this regression and getting the fix into the 
release is something the release management team needs to look into. At 
a minimum, this issue being fixed on the 22nd should have prompted an 
-rc2 release, with this issue being listed as a 'blocker' for the 
eventual 1.6.2.15 release. In fact, this issue was known about for about 
three weeks before 1.6.2.15-rc1 was made, so I'd suggest that the -rc 
shouldn't have even been made with this outstanding. There was a 
breakdown in the process somewhere.

-- 
Kevin P. Fleming
Digium, Inc. | Director of Software Technologies
445 Jan Davis Drive NW - Huntsville, AL 35806 - USA
skype: kpfleming | jabber: kfleming at digium.com
Check us out at www.digium.com & www.asterisk.org