[
https://issues.apache.org/jira/browse/HADOOP-2145?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12540290
]
Doug Cutting commented on HADOOP-2145:
--------------------------------------
We might not put the libs in hadoop/trunk/lib, but instead in hadoop/site/lib.
The build.xml could refer to them as "../site/lib", which would work if you'd
checked out the site stuff too, which most folks who're working on
documentation should. Would that work?
FYI, I assume we intend to restructure things like:
{noformat}
hadoop/
branches/
tags/
trunk/
site/
author/
publish/
{noformat}
So that people.apache.org:/www/lucene.apache.org/ then has the following things
checked out from subversion to build the site:
{noformat}
hadoop/ -- http://svn.apache.org/repos/asf/lucene/hadoop/site
0.15.0/ --
http://svn.apache.org/repos/asf/lucene/hadoop/tags/release-0.15.0/docs
0.14.3/ --
http://svn.apache.org/repos/asf/lucene/hadoop/tags/release-0.14.3/docs
{noformat}
The site pages would then contain links to the version-specific docs. All user
documentation would go in version-specific docs, and the site would mostly be a
shell, with some project & developer information and maybe a very high-level
introduction. All the details would be pushed into the versioned stuff.
Do we have an issue in Jira yet for this restructuring of the docs? If not, we
should add one.
Note that a side-effect of this would be that jira-issues related to the site
(but not the docs) are not version-specific: they'd have no fix-version or
affects-version.
> need 'doc' target that runs forrest
> -----------------------------------
>
> Key: HADOOP-2145
> URL: https://issues.apache.org/jira/browse/HADOOP-2145
> Project: Hadoop
> Issue Type: New Feature
> Components: build
> Reporter: Doug Cutting
> Assignee: Raghu Angadi
> Fix For: 0.16.0
>
>
> We should have an 'ant doc' target that runs forrest. Ideally we'd check the
> forrest jars into subversion, so that this task works out of the box, w/o
> having to separately install forrest. The ant task could both build the docs
> and copy them into the docs/ directory, rather than requiring this to be done
> manually.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
