Why did we accept a patch that fundamentally changes the user
experience without requiring that the patch included updated
documentation in forest of javadoc?
I propose that committers should reject feature patches that do not
included documentation.
Thoughts?
---
On a related note, does the release check for mapred-default.xml and
warn if it exists? Seems like a good idea.
On Sep 14, 2007, at 9:23 PM, Arun C Murthy wrote:
Please note:
If, and only if, you're running Hadoop from straight from the svn
*trunk*, as of Sept 12 Hadoop no longer uses the *mapred-
default.xml* file.
Please move your mapred-default.xml configuration properties to
your *hadoop-site.xml*. Hadoop's configuration also supports the
notion of *final* parameters which, once specified, cannot be
overridden by subsequently loaded configuration files. These
changes will roll-out in the hadoop-0.15.0 release only.
Currently, the only documentation on how the new configuration
works is at http://issues.apache.org/jira/browse/HADOOP-785. Wiki
(or other) documentation, is forthcoming (http://issues.apache.org/
jira/browse/HADOOP-1881).
thanks,
Arun
