[asterisk-users] Upgrade to Asterisk 1.4 - it's one year's old!

[Tilghman Lesher]

On Saturday 15 December 2007 10:02:23 Rob Hillis wrote:
> One of the biggest barriers to upgrading are the number of little
> "gotchas" in syntax changes that can make an upgrade from 1.2 to 1.4
> quite painful.  After the pain I went through upgrading to 1.4, I've
> always been recommending to people to think twice about upgrading if 1.2
> does what they require.
>
> Many of the changes may have been seen as minor - one or two changes are
> to be expected, but I ran into at least half a dozen - mostly variable
> changes if I recall correctly - things such as deprecating CALLERIDNUM
> in favour of CALLERID(num).  Some of the breakage was minor (e.g. loss
> of caller ID processing) but some of them resulted in calls being
> dropped in unpredictable places.
>
> All I can say is with 1.6, if a change is made that causes something
> that worked in 1.4 not to work in 1.6, please think twice, three times
> or four times before making the change, or making the change in such a
> way that it won't break dialplan stuff from 1.4.

If anything broke from the transition from 1.2 to 1.4, it is because you were
using something that was deprecated in 1.2.  What we had attempted to do
in deprecation modes was to print the warning ONCE for each deprecated
operation, per Asterisk startup.  I think that this was much too conservative.
It is very easy to miss that deprecation warning, since it occurs so few
times.  Of course, the opposite side is that we don't want deprecation
warnings to fill up your logs, so there's a balancing act here.  But we could
probably do with making the deprecation warnings a bit more prominent
and print them multiple times (for example, every 10th usage).  That should
make it more clear that there's something to change.

Of course, all of these deprecations should be covered in UPGRADE.txt, so
please read that file every time you upgrade to a new version.  It will
contain everything that has changed in a possibly incompatible way.  And if
you find something that broke that wasn't in this file, please let us know, so
we can revise that file.  We may not have gotten everything, but we do try.

-- 
Tilghman