The issue is maintainability of the River code. Many of the changes
between 1.4 and 1.6 are not about making it possible to do something
that could not be done in 1.4, but about making the code simpler and
cleaner, and reducing the risk of errors. The problem with the "requires
a compelling reason" view is that no one maintainability issue, by
itself, is a deal killer. What kills projects is the cumulative effect
of many, many small maintainability issues.
There is also a testing issue. We should test with each version we claim
to support. At the last count, a full QA test takes about 20 hours per
version.
I think there is a non-trivial trade-off between the cost of supporting
1.4 or 1.5 and the cost of not supporting 1.4 or 1.5. To resolve it we
need feedback on the extent to which users really are depending on 1.4
and 1.5 run time support.
Patricia
On 9/17/2010 5:24 AM, Dennis Brake wrote:
Since you can always compile on one version and run on a higher
version, the core software should always be built with the lowest
version that contains all required capabilities.
When moving from one version to a higher version requires a compelling
reason. Not just ' because its the newest.'
Dennis
Swyped on my android
----- Reply message -----
From: "Tom Hobbs"<[email protected]>
Date: Fri, Sep 17, 2010 7:06 am
Subject: JDK Version Migration
To: "[email protected]"<[email protected]>
Dear All,
There is talk on the River developer list regarding which JDK version River
is currently compiled and distributed under.
Current it's released under JDK 1.4, however issues around moving toward JDK
5 or 6 are underway. Obviously such a move would have consequences in many
different environments, particularly when moving from one version of the
River to the next.
Do any of the users subscribed to this list have any views on the upgrade
path and how far along that path (in terms of Java version and target dates)
River should go?
Many thanks for your time,
Tom
