forrest-svn mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From cross...@apache.org
Subject svn commit: r233290 - /forrest/trunk/site-author/content/xdocs/guidelines.xml
Date Thu, 18 Aug 2005 06:25:05 GMT
Author: crossley
Date: Wed Aug 17 23:25:01 2005
New Revision: 233290

URL: http://svn.apache.org/viewcvs?rev=233290&view=rev
Log:
Expand the "Code management" section, clarify the commit-then-review
policy for decision-making, provide some tips for patch application.

Modified:
    forrest/trunk/site-author/content/xdocs/guidelines.xml

Modified: forrest/trunk/site-author/content/xdocs/guidelines.xml
URL: http://svn.apache.org/viewcvs/forrest/trunk/site-author/content/xdocs/guidelines.xml?rev=233290&r1=233289&r2=233290&view=diff
==============================================================================
--- forrest/trunk/site-author/content/xdocs/guidelines.xml (original)
+++ forrest/trunk/site-author/content/xdocs/guidelines.xml Wed Aug 17 23:25:01 2005
@@ -518,15 +518,58 @@
 
   <section id="code">
     <title>Code management</title>
+
+    <p>
+      The term "patch" has two meanings: Developers provide a set of changes
+      via our <link href="site:bugs">Issue Tracker</link> marked for inclusion,
+      which will be applied by a committer. Committers apply their own work
+      directly, but it is still essentially a patch.
+    </p>
+
     <p>
       We use the
       <link href="http://www.apache.org/foundation/glossary.html#CommitThenReview">Commit-then-review</link>
-      method. This basically means that committers conduct a very minor
-      review of each patch, looking for major issues such as licensing
-      issues and obvious things that would break the build of trunk.
-      Otherwise they would add the patch as-is, then everyone will
-      review the changes.
+      method for decision-making about code changes. Please refer to that
+      glossary definition. Note that it does not preclude the committer
+      from making changes to patches prior to their commit, nor mean that the
+      committer can be careless. Rather it is a policy for decision-making.
+    </p>
+
+    <p>
+      There is an important issue where both developers and committers need
+      to pay special attention: "licenses". We must not introduce licensing
+      conditions that go beyond the terms of the <link href="ext:asl">Apache License</link>.
+      If such issues do creep in to our repository, then we must work as
+      quickly as possible to address it and definitely before the next release.
+    </p>
+
+    <p>
+      There are some other problem areas:
+      What should a committer do if the patch is sloppy, containing inconsistent
+      whitespace and other code formatting, which mean that actual changes are not
+      easily visible in the svn diff messages. What if there is poorly constructed
+      (yet working) xml or java code? What if the new functionality is beyond the
+      scope of the project? What if there is a better way to do the task? What if
+      the patch will break the build, thereby preventing other developers from
+      working and causing an unstable trunk?
+    </p>
+
+    <p>
+      The committer has various options: ask the developer to resubmit the patch;
+      change the patch to fix the problems prior to committing; discuss the
+      issues on the dev list; commit it and then draw attention to the
+      issues so that the rest of the community can review and fix it.
+      A combination of these options would appear to be the best approach.
+      Please aim to not break the build, or introduce license problems, or make
+      noisy changes that obscure the real differences.
+    </p>
+
+    <p>
+      Committers should not be afraid to add changes that still need attention.
+      This enables prompt patch application and eases the load on the individual
+      committer. An interesting side-effect is that it encourages community growth.
     </p>
+
   </section>
 
 <!-- FIXME:



Mime
View raw message