Hi Allen. Thank you for putting together this documentation.
For shelldocmd.py, what Python version did you use in testing? I don't know for sure that it's still the case, but I recall that fairly recent CentOS versions are pinned to Python 2.6 as part of the implementation of yum. I'd like to suggest that we stick to base 2.6 with no additional modules required. (I don't see any red flags so far.)
UnixShellGuide.md doesn't appear to be hyperlinked from anywhere else in the documentation. Shall we add it to the left nav?
Shall we omit all private non-replaceable functions from the documentation? On the Java side, we filter out the private things. It's still helpful for maintainers to have the full docs on these functions in the source code, but I don't think we need to publish it to end users.
hadoop_add_colonpath probably needs more context in the description. We could mention that the default implementations of hadoop_add_javalibpath and hadoop_add_ldlibpath use this. It doesn't currently mention that the first argument is the name of the out variable to receive the modification. Alternatively, I think we could declare this one as a private non-replaceable implementation detail and filter it out of the docs as per above comment.
Here are a few typos I spotted:
- is controlled via [the shell](CommandsManula.html) (The hyperlink target should be CommandsManual.html.)
- HADOOP_CLIENT_OPTS="-Xmx1g -Dhadoop.socks.server=localhost:4000 hadoop fs -ls /tmp" (I think you meant to close the double-quote before the start of the hadoop command.)
- to run hadoop commands access the server (I think this was supposed to be "accessing".)
- In the .hadooprc example, the if statement has an extra closing curly brace around HADOOP_SERVER.
- There are many enironment variables (It should be "environment".)
- the series of `_OPT` variables (I think this was supposed to be OPTS.)
- Advanced administrators may which to supplement (This should be "may wish".)
- provides the capabilities to do funcion overrides (This should be "function".)
- Different places in the doc say either "run time" or "runtime". Let's use one consistently, probably "runtime".
- Hadoop's shell code has a [function library](./HadoopShellFunctionAPI.html) (The hyperlink target doesn't match the actual file name, which is UnixShellAPI.html.)
- Print a message to stderr if –debug is turuned on (This should use "turned on".)