Merged revisions 2136-2200,2202-2290,2292-2301 via svnmerge from

http://scons.tigris.org/svn/scons/branches/core

........
  r2145 | stevenknight | 2007-07-17 09:15:12 -0500 (Tue, 17 Jul 2007) | 3 lines

  Don't put null strings (from variable expansion) in a path list.
  (They get turned into the current directory on later expansion.)
........
  r2146 | stevenknight | 2007-07-17 10:47:39 -0500 (Tue, 17 Jul 2007) | 3 lines

  Add support for optional arguments on command-line long options
  by specifying nargs='?'.
........
  r2149 | stevenknight | 2007-07-17 15:22:24 -0500 (Tue, 17 Jul 2007) | 2 lines

  Remove left-over Optik mentions.
........
  r2150 | stevenknight | 2007-07-17 15:39:34 -0500 (Tue, 17 Jul 2007) | 4 lines

  Add a $SWIGPATH variable for finding SWIG dependencies, with
  $SWIGINC{PREFIX,SUFFIX} for adding them to the command line.
........
  r2154 | stevenknight | 2007-07-18 20:05:31 -0500 (Wed, 18 Jul 2007) | 2 lines

  Fix variable misspellings in the doc added for $SWIGOUTPUT.
........
  r2155 | stevenknight | 2007-07-18 20:07:28 -0500 (Wed, 18 Jul 2007) | 2 lines

  Add the Python eggs info file to the RPM packaging build.
........
  r2156 | stevenknight | 2007-07-18 20:15:08 -0500 (Wed, 18 Jul 2007) | 2 lines

  Convert documentation from DocBook SGML to XML.
........
  r2158 | stevenknight | 2007-07-19 17:16:19 -0500 (Thu, 19 Jul 2007) | 3 lines

  Conditionally add the .egg-info the RPM file list only if the distutils
  in the version of Python that rpmbuild will execute knows about them.
........
  r2161 | stevenknight | 2007-07-19 19:12:29 -0500 (Thu, 19 Jul 2007) | 5 lines

  Capture a test case (contributed by Tilo Prutz) where instantiation of
  a private class causes javac to generate an additional anonymous inner
  class file.  (No solution yet, but there's no sense throwing away the
  preparatory work.)
........
  r2162 | stevenknight | 2007-07-20 11:29:56 -0500 (Fri, 20 Jul 2007) | 3 lines

  Support passing a list of .java files as source to the Java() builder.
  (Leanid Nazdrynau)
........
  r2163 | garyo | 2007-07-20 12:00:35 -0500 (Fri, 20 Jul 2007) | 1 line

  Fixed cut-n-paste error in Touch factory method doc in users guide.
........
  r2167 | stevenknight | 2007-07-21 22:59:40 -0500 (Sat, 21 Jul 2007) | 2 lines

  Don't execute the SWIGOUTDIR test if swig isn't installed.
........
  r2168 | stevenknight | 2007-07-21 23:14:17 -0500 (Sat, 21 Jul 2007) | 2 lines

  Fix the test's ability to run under a path name containing spaces.
........
  r2171 | stevenknight | 2007-07-24 15:54:41 -0500 (Tue, 24 Jul 2007) | 2 lines

  Handle white space in key file names in the packaging build.
........
  r2172 | stevenknight | 2007-07-24 21:41:15 -0500 (Tue, 24 Jul 2007) | 2 lines

  More efficient copying of construction environments.
........
  r2173 | stevenknight | 2007-07-25 10:56:02 -0500 (Wed, 25 Jul 2007) | 2 lines

  Update the SCons build for Subversion and general clean-up.
........
  r2174 | stevenknight | 2007-07-25 11:35:16 -0500 (Wed, 25 Jul 2007) | 3 lines

  Suppress the [brackets] around a node in the --tree=prune output if
  the node is a source.
........
  r2175 | stevenknight | 2007-07-25 12:52:18 -0500 (Wed, 25 Jul 2007) | 3 lines

  Commonize the skip_test() method and make its behavior configurable
  via a TESTCOMMON_PASS_SKIPS environment variable.
........
  r2178 | stevenknight | 2007-07-25 21:43:47 -0500 (Wed, 25 Jul 2007) | 3 lines

  Add $JAVACLASSPATH and $JAVASOURCEPATH construction variables.  (Leanid
  Nazdrynau)
........
  r2182 | stevenknight | 2007-07-30 12:10:20 -0500 (Mon, 30 Jul 2007) | 3 lines

  Refactor Builder suffix-adjusting into its own method, so we can
  (potentially) re-use it for Builders with attached source Builders.
........
  r2183 | stevenknight | 2007-07-30 14:51:53 -0500 (Mon, 30 Jul 2007) | 2 lines

  More efficient source-builder suffix matching.
........
  r2184 | stevenknight | 2007-07-30 16:01:42 -0500 (Mon, 30 Jul 2007) | 4 lines

  Encapsulate initialization of the default FS object by an accessor
  function in SCons.Node.FS.  (This also gets rid of an unnecessary
  reference to SCons.Node.FS.default_fs in the LaTeX scanner.)
........
  r2193 | stevenknight | 2007-07-30 18:24:07 -0500 (Mon, 30 Jul 2007) | 3 lines

  Fix interpretation of source arguments that have no suffix when the
  called Builder has both a src_suffix and a src_builder.
........
  r2194 | stevenknight | 2007-07-31 10:25:31 -0500 (Tue, 31 Jul 2007) | 2 lines

  Increase the number of tries for random output from three to ten.
........
  r2195 | stevenknight | 2007-07-31 10:52:28 -0500 (Tue, 31 Jul 2007) | 3 lines

  Skip the test gracefully if the zipfile module can't read the file it
  just wrote (which is the case for Python 2.1 on 64-bit systems).
........
  r2196 | stevenknight | 2007-07-31 13:06:21 -0500 (Tue, 31 Jul 2007) | 2 lines

  Move the "import zipfile" so it doesn't fail on Python <= 2.0.
........
  r2197 | stevenknight | 2007-07-31 14:51:50 -0500 (Tue, 31 Jul 2007) | 3 lines

  Commonize initialization of the various Java builders so they can be
  hooked up into a multi-stage Builder chain.  (Leanid Nazdrynau)
........
  r2198 | stevenknight | 2007-07-31 16:15:18 -0500 (Tue, 31 Jul 2007) | 3 lines

  Fix use of ${TARGET.dir} and ${SOURCE.dir} expansions in $FORTRANMODDIR
  $JARCHDIR, $JARFLAGS, $LEXFLAGS, $SWIGFLAGS, $SWIGOUTDIR and $YACCFLAGS.
........
  r2199 | stevenknight | 2007-07-31 16:25:48 -0500 (Tue, 31 Jul 2007) | 2 lines

  Remove left-over Trace() call.
........
  r2202 | stevenknight | 2007-08-01 12:31:48 -0500 (Wed, 01 Aug 2007) | 3 lines

  Bail out via test.skip_test() if wix ("candle") isn't found.
  Put the main body of code flush left instead of under an if: block.
........
  r2203 | stevenknight | 2007-08-01 15:35:55 -0500 (Wed, 01 Aug 2007) | 5 lines

  Fix Tool.packaging.rpm.package() so it doesn't always overwrite
  $RPMFLAGS with -ta.
  Set --buildroot in RPM packaging tests so they don't overwrite
  each other when run simultaneously.
........
  r2204 | stevenknight | 2007-08-01 15:37:36 -0500 (Wed, 01 Aug 2007) | 2 lines

  Fix a nested scope issue with the internal build_sources() function.
........
  r2205 | stevenknight | 2007-08-01 15:46:08 -0500 (Wed, 01 Aug 2007) | 5 lines

  Normalize (X out) the CreationDate field inside embedded, compressed
  PostScript streams within the generated PDF files.  Also normalize
  preceding Length field, since compression length is affected by different
  patterns of input, including the variable CreationDate value.
........
  r2211 | stevenknight | 2007-08-02 08:52:06 -0500 (Thu, 02 Aug 2007) | 2 lines

  Add the new modules from branches/packaging to the SCons packaging build.
........
  r2212 | stevenknight | 2007-08-02 19:59:01 -0500 (Thu, 02 Aug 2007) | 2 lines

  Fix the JAVACLASSPATH test when javah isn't on the default $PATH.
........
  r2214 | stevenknight | 2007-08-03 15:05:21 -0500 (Fri, 03 Aug 2007) | 4 lines

  Hook up the Java builders into a multi-step chain underneath a Java()
  pseudo-builder (wrapper) that examines its arguments and calls the
  appropriate underlying file-or-dir builder.
........
  r2215 | stevenknight | 2007-08-03 15:49:58 -0500 (Fri, 03 Aug 2007) | 2 lines

  Fix for old Python versions:  use apply() instead of *args, **kw.
........
  r2216 | stevenknight | 2007-08-03 16:49:31 -0500 (Fri, 03 Aug 2007) | 2 lines

  Hook up the SWIG builder as a source builder for .java files.
........
  r2217 | stevenknight | 2007-08-03 17:28:19 -0500 (Fri, 03 Aug 2007) | 2 lines

  Don't use .endswith(), which didn't appear until later Python versions.
........
  r2218 | stevenknight | 2007-08-03 17:29:38 -0500 (Fri, 03 Aug 2007) | 2 lines

  Replace tabs with spaces.
........
  r2219 | stevenknight | 2007-08-04 08:06:23 -0500 (Sat, 04 Aug 2007) | 3 lines

  Initialize a loop-invariant lambda for matching .java suffixes outside
  the loop.
........
  r2220 | stevenknight | 2007-08-07 15:06:13 -0500 (Tue, 07 Aug 2007) | 2 lines

  Refactor parallel class-generation loops into one.
........
  r2221 | stevenknight | 2007-08-07 16:04:06 -0500 (Tue, 07 Aug 2007) | 5 lines

  Have the Java multi-step builder test actually check for generated files,
  and fix the generation of .java and .class file names, and interaction
  with the SWIG builder, so that the files are generated in the correct
  place.
........
  r2222 | stevenknight | 2007-08-07 16:45:05 -0500 (Tue, 07 Aug 2007) | 3 lines

  Fix dependencies on SWIG-generated .java files so they don't have to
  be built in multiple passes.
........
  r2226 | stevenknight | 2007-08-07 18:00:22 -0500 (Tue, 07 Aug 2007) | 2 lines

  Fix SWIG when used with BuildDir().
........
  r2227 | stevenknight | 2007-08-07 22:15:55 -0500 (Tue, 07 Aug 2007) | 5 lines

  User's guide updates:
  - Make the multiple files example match its text.
  - Expand a truncated sentence about being able to use Python function actions
    in the Command() Builder.
........
  r2228 | stevenknight | 2007-08-07 23:25:18 -0500 (Tue, 07 Aug 2007) | 3 lines

  Don't generate an error if a #include file matches a same-named
  directory in $CPPPATH (or $FORTRANPATH, etc.).
........
  r2229 | stevenknight | 2007-08-07 23:40:00 -0500 (Tue, 07 Aug 2007) | 2 lines

  Fix a code example.  (Gary Oberbrunner)
........
  r2230 | stevenknight | 2007-08-08 00:05:43 -0500 (Wed, 08 Aug 2007) | 3 lines

  Capture a test case to make sure AddPostAction() doesn't interfere
  with normal linking.  (Matt Doar, Gary Oberbrunner)
........
  r2233 | stevenknight | 2007-08-08 14:15:44 -0500 (Wed, 08 Aug 2007) | 2 lines

  Fix documentation typo in a construction variable cross-reference.
........
  r2234 | stevenknight | 2007-08-08 17:03:25 -0500 (Wed, 08 Aug 2007) | 2 lines

  Changes to SCons packaging to support checkpoint releases.
........
  r2235 | stevenknight | 2007-08-09 10:10:01 -0500 (Thu, 09 Aug 2007) | 2 lines

  Sidestep false negatives on heavily loaded systems.
........
  r2236 | garyo | 2007-08-09 11:16:26 -0500 (Thu, 09 Aug 2007) | 1 line

  Allow unpackaged files (e.g. *.pyo) to exist in the build dir without being packaged in the RPM.  Without this, on some systems the rpmbuild may error out.
........
  r2237 | stevenknight | 2007-08-09 11:27:56 -0500 (Thu, 09 Aug 2007) | 5 lines

  Fix test/SWIG/build-dir.py so it works on old Python versions without
  distutils.sysconfig.
  Instead of just cutting-and-pasting initialization code from other
  SWIG tests, centralize it in some new TestSCons methods.
........
  r2238 | garyo | 2007-08-09 11:30:58 -0500 (Thu, 09 Aug 2007) | 1 line

  Use docbook 4.3 instead of 4.4 for the XML doctype since some older(?) jade parsers can't handle new 4-byte Unicode chars in the 4.4 version of isogrk4.ent.
........
  r2240 | stevenknight | 2007-08-09 16:35:06 -0500 (Thu, 09 Aug 2007) | 2 lines

  User's Guide updates (post packaging changes).
........
  r2243 | stevenknight | 2007-08-10 10:31:51 -0500 (Fri, 10 Aug 2007) | 3 lines

  Fix the User's Guide build to use openjade, and to accomodate a change
  in the name of the main generated file (book1.html => index.html).
........
  r2245 | stevenknight | 2007-08-10 11:09:16 -0500 (Fri, 10 Aug 2007) | 2 lines

  Update the {CHANGES,RELEASE}.txt datestamp lines.
........
  r2253 | stevenknight | 2007-08-10 16:21:54 -0500 (Fri, 10 Aug 2007) | 2 lines

  Fix the wix Tool module's ability to handle null entries in $PATH.
........
  r2261 | stevenknight | 2007-08-11 23:08:12 -0500 (Sat, 11 Aug 2007) | 3 lines

  Remove unnecessary files (.svnt/*, .{ae,cvs}ignore, www/*) from the
  scons-src packages.
........
  r2262 | stevenknight | 2007-08-11 23:24:49 -0500 (Sat, 11 Aug 2007) | 2 lines

  Add missing __revision__ lines.
........
  r2263 | stevenknight | 2007-08-11 23:33:42 -0500 (Sat, 11 Aug 2007) | 2 lines

  Skip the test if the MANIFEST file hasn't been built.
........
  r2264 | stevenknight | 2007-08-11 23:36:30 -0500 (Sat, 11 Aug 2007) | 2 lines

  Add recent compatibility modules to the relevant exceptions lists.
........
  r2265 | stevenknight | 2007-08-11 23:39:00 -0500 (Sat, 11 Aug 2007) | 3 lines

  Update __VERSION__ strings in the QMTest/*.py modules, so that packaging
  tests (src/test_*.py) will pass after builds of checkpoint releases.
........
  r2266 | stevenknight | 2007-08-12 07:36:19 -0500 (Sun, 12 Aug 2007) | 2 lines

  Add a comment about why we construct the __VERSION__ string at run time.
........
  r2267 | stevenknight | 2007-08-12 07:42:30 -0500 (Sun, 12 Aug 2007) | 2 lines

  Avoid reading the MANIFEST file twice.  (Courtesy review by Greg Noel.)
........
  r2268 | stevenknight | 2007-08-12 08:14:53 -0500 (Sun, 12 Aug 2007) | 3 lines

  Shift Install() and InstallAs() from being documented as functions
  to being documented as Builders.
........
  r2269 | garyo | 2007-08-13 08:49:52 -0500 (Mon, 13 Aug 2007) | 1 line

  Tests: Skip some more Java tests if javac is not installed on the test machine so they don't get marked as failing.
........
  r2270 | garyo | 2007-08-13 11:09:39 -0500 (Mon, 13 Aug 2007) | 1 line

  Fixed typo in test (shows up on non-Linux platforms).
........
  r2271 | garyo | 2007-08-13 14:09:05 -0500 (Mon, 13 Aug 2007) | 4 lines

  Test portability fixes for Darwin/OSX and IRIX.
  This does not make all the tests pass on those OSes,
  but it takes care of some of the more obvious errors that
  I have time for right now.  More to come.
........
  r2272 | stevenknight | 2007-08-13 15:33:29 -0500 (Mon, 13 Aug 2007) | 2 lines

  Tab => space fix.
........
  r2273 | stevenknight | 2007-08-13 15:33:52 -0500 (Mon, 13 Aug 2007) | 2 lines

  Test for swig, too, which is used to build from the .i file.
........
  r2277 | garyo | 2007-08-14 10:40:00 -0500 (Tue, 14 Aug 2007) | 8 lines

  Test portability on IRIX: test/Actions/pre-post creates target file
  before building target, then IRIX CC does not chmod +x afterwards.
  I think this change is safe on all OSes.

  test/AS/ml.py: I think this is only supposed to be run on win32
  (not skipped only on win32); the sense of the skip test was backwards.
........
  r2278 | stevenknight | 2007-08-14 11:04:40 -0500 (Tue, 14 Aug 2007) | 2 lines

  Add -tt when running tests, to catch inconsistent tab usage.
........
  r2279 | stevenknight | 2007-08-14 14:00:43 -0500 (Tue, 14 Aug 2007) | 2 lines

  Minor refactor of logic in File.retrieve_from_cache().
........
  r2280 | stevenknight | 2007-08-15 01:11:40 -0500 (Wed, 15 Aug 2007) | 2 lines

  Refactor CacheDir support into its own module.
........
  r2281 | stevenknight | 2007-08-15 07:24:51 -0500 (Wed, 15 Aug 2007) | 2 lines

  Move the cachepath() method from FS.File to the CacheDir class.
........
  r2282 | stevenknight | 2007-08-15 08:31:34 -0500 (Wed, 15 Aug 2007) | 2 lines

  Python 1.5.2 fix in the new Null class.
........
  r2283 | stevenknight | 2007-08-15 10:45:53 -0500 (Wed, 15 Aug 2007) | 5 lines

  Refactor CacheDir unit tests to:
  - restore functionality that was dropped in the transition;
  - commonize creation of test Nodes and other (mock) objects
  - separate CacheDir tests from tests of CacheDir through Node.FS.File.
........
  r2284 | stevenknight | 2007-08-15 11:46:38 -0500 (Wed, 15 Aug 2007) | 3 lines

  Replace the Executor.Null.NullEnvironment object with a real Null object,
  so it will absorb the CacheDir method calls as well.
........
  r2285 | stevenknight | 2007-08-15 11:52:57 -0500 (Wed, 15 Aug 2007) | 5 lines

  Add a get_CacheDir() method to a construction environment, which will
  be used to fetch per-environment CacheDir specifications.  (Right now
  all calls to it still just return the one attached to underlying default
  FS object.)
........
  r2286 | stevenknight | 2007-08-15 15:15:46 -0500 (Wed, 15 Aug 2007) | 2 lines

  Support per-construction-environment configuration of CacheDir().
........
  r2287 | stevenknight | 2007-08-15 15:33:04 -0500 (Wed, 15 Aug 2007) | 2 lines

  Move the tests of CacheDir()-related command-line options into test/CacheDir.
........
  r2293 | stevenknight | 2007-08-16 11:14:49 -0500 (Thu, 16 Aug 2007) | 3 lines

  Add the Package() builder description to the documentation build,
  fixing the XML so that it will build.
........
  r2294 | stevenknight | 2007-08-16 12:51:19 -0500 (Thu, 16 Aug 2007) | 3 lines

  Reorganize packaging documentation:  alphabetize the variable definitions
  (and function names), document Tag() as a function, not a builder.
........
  r2296 | stevenknight | 2007-08-16 12:55:01 -0500 (Thu, 16 Aug 2007) | 2 lines

  Add a build command.
........
  r2300 | stevenknight | 2007-08-16 16:49:13 -0500 (Thu, 16 Aug 2007) | 2 lines

  First cut at documenting packaging variables.
........
  r2301 | stevenknight | 2007-08-16 16:51:21 -0500 (Thu, 16 Aug 2007) | 3 lines

  Construct the .src.rpm and .arch.rpm file names independnetly, not
  by trying to massage one into the other.
........

1 files changed, 492 insertions, 0 deletions

diff --git a/doc/user/caching.xml b/doc/user/caching.xml
new file mode 100644
index 0000000..51b30ae
--- /dev/null
+++ b/doc/user/caching.xml
@@ -0,0 +1,492 @@
+<!--
+
+  __COPYRIGHT__
+
+  Permission is hereby granted, free of charge, to any person obtaining
+  a copy of this software and associated documentation files (the
+  "Software"), to deal in the Software without restriction, including
+  without limitation the rights to use, copy, modify, merge, publish,
+  distribute, sublicense, and/or sell copies of the Software, and to
+  permit persons to whom the Software is furnished to do so, subject to
+  the following conditions:
+
+  The above copyright notice and this permission notice shall be included
+  in all copies or substantial portions of the Software.
+
+  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY
+  KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE
+  WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+  NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
+  LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+  OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+  WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+-->
+
+  <para>
+
+  On multi-developer software projects,
+  you can sometimes speed up every developer's builds a lot by
+  allowing them to share the derived files that they build.
+  &SCons; makes this easy, as well as reliable.
+
+  </para>
+
+  <section>
+  <title>Specifying the Shared Cache Directory</title>
+
+    <para>
+
+    To enable sharing of derived files,
+    use the &CacheDir; function
+    in any &SConscript; file:
+
+    </para>
+
+    <programlisting>
+       CacheDir('/usr/local/build_cache')
+    </programlisting>
+
+    <para>
+
+    Note that the directory you specify must already exist
+    and be readable and writable by all developers
+    who will be sharing derived files.
+    It should also be in some central location
+    that all builds will be able to access.
+    In environments where developers are using separate systems
+    (like individual workstations) for builds,
+    this directory would typically be
+    on a shared or NFS-mounted file system.
+
+    </para>
+
+    <para>
+
+    Here's what happens:
+    When a build has a &CacheDir; specified,
+    every time a file is built,
+    it is stored in the shared cache directory
+    along with its MD5 build signature.
+      <footnote>
+      <para>
+      Actually, the MD5 signature is used as the name of the file
+      in the shared cache directory in which the contents are stored.
+      </para>
+      </footnote>
+    On subsequent builds,
+    before an action is invoked to build a file,
+    &SCons; will check the shared cache directory
+    to see if a file with the exact same build
+    signature already exists.
+    If so, the derived file will not be built locally,
+    but will be copied into the local build directory
+    from the shared cache directory,
+    like so:
+
+    </para>
+
+    <screen>
+      % <userinput>scons -Q</userinput>
+      cc -o hello.o -c hello.c
+      cc -o hello hello.o
+      % <userinput>scons -Q -c</userinput>
+      Removed hello.o
+      Removed hello
+      % <userinput>scons -Q</userinput>
+      Retrieved `hello.o' from cache
+      Retrieved `hello' from cache
+    </screen>
+
+  </section>
+
+  <section>
+  <title>Keeping Build Output Consistent</title>
+
+    <para>
+
+    One potential drawback to using a shared cache
+    is that the output printed by &SCons;
+    can be inconsistent from invocation to invocation,
+    because any given file may be rebuilt one time
+    and retrieved from the shared cache the next time.
+    This can make analyzing build output more difficult,
+    especially for automated scripts that
+    expect consistent output each time.
+
+    </para>
+
+    <para>
+
+    If, however, you use the <literal>--cache-show</literal> option,
+    &SCons; will print the command line that it
+    <emphasis>would</emphasis> have executed
+    to build the file,
+    even when it is retrieving the file from the shared cache.
+    This makes the build output consistent
+    every time the build is run:
+
+    </para>
+
+    <screen>
+      % <userinput>scons -Q</userinput>
+      cc -o hello.o -c hello.c
+      cc -o hello hello.o
+      % <userinput>scons -Q -c</userinput>
+      Removed hello.o
+      Removed hello
+      % <userinput>scons -Q --cache-show</userinput>
+      cc -o hello.o -c hello.c
+      cc -o hello hello.o
+    </screen>
+
+    <para>
+
+    The trade-off, of course, is that you no longer
+    know whether or not &SCons;
+    has retrieved a derived file from cache
+    or has rebuilt it locally.
+
+    </para>
+
+  </section>
+
+  <section>
+  <title>Not Using the Shared Cache for Specific Files</title>
+
+    <para>
+
+    You may want to disable caching for certain
+    specific files in your configuration.
+    For example, if you only want to put
+    executable files in a central cache,
+    but not the intermediate object files,
+    you can use the &NoCache;
+    function to specify that the
+    object files should not be cached:
+
+    </para>
+
+    <programlisting>
+       env = Environment()
+       obj = env.Object('hello.c')
+       env.Program('hello.c')
+       CacheDir('cache')
+       NoCache('hello.o')
+    </programlisting>
+
+    <para>
+
+    Then when you run &scons; after cleaning
+    the built targets,
+    it will recompile the object file locally
+    (since it doesn't exist in the shared cache directory),
+    but still realize that the shared cache directory
+    contains an up-to-date executable program
+    that can be retrieved instead of re-linking:
+
+    </para>
+
+    <!--
+
+    <scons_output example="ex1">
+      <scons_output_command>scons -Q</scons_output_command>
+      <scons_output_command>scons -Q -c</scons_output_command>
+      <scons_output_command>scons -Q</scons_output_command>
+    </scons_output>
+
+    -->
+
+    <screen>
+      % <userinput>scons -Q</userinput>
+      cc -o hello.o -c hello.c
+      cc -o hello hello.o
+      % <userinput>scons -Q -c</userinput>
+      Removed hello.o
+      Removed hello
+      % <userinput>scons -Q</userinput>
+      cc -o hello.o -c hello.c
+      Retrieved `hello' from cache
+    </screen>
+
+  </section>
+
+  <section>
+  <title>Disabling the Shared Cache</title>
+
+    <para>
+
+    Retrieving an already-built file
+    from the shared cache
+    is usually a significant time-savings
+    over rebuilding the file,
+    but how much of a savings
+    (or even whether it saves time at all)
+    can depend a great deal on your
+    system or network configuration.
+    For example, retrieving cached files
+    from a busy server over a busy network
+    might end up being slower than
+    rebuilding the files locally.
+
+    </para>
+
+    <para>
+
+    In these cases, you can specify
+    the <literal>--cache-disable</literal>
+    command-line option to tell &SCons;
+    to not retrieve already-built files from the
+    shared cache directory:
+
+    </para>
+
+    <screen>
+      % <userinput>scons -Q</userinput>
+      cc -o hello.o -c hello.c
+      cc -o hello hello.o
+      % <userinput>scons -Q -c</userinput>
+      Removed hello.o
+      Removed hello
+      % <userinput>scons -Q</userinput>
+      Retrieved `hello.o' from cache
+      Retrieved `hello' from cache
+      % <userinput>scons -Q -c</userinput>
+      Removed hello.o
+      Removed hello
+      % <userinput>scons -Q --cache-disable</userinput>
+      cc -o hello.o -c hello.c
+      cc -o hello hello.o
+    </screen>
+
+  </section>
+
+  <section>
+  <title>Populating a Shared Cache With Already-Built Files</title>
+
+    <para>
+
+    Sometimes, you may have one or more derived files
+    already built in your local build tree
+    that you wish to make available to other people doing builds.
+    For example, you may find it more effective to perform
+    integration builds with the cache disabled
+    (per the previous section)
+    and only populate the shared cache directory
+    with the built files after the integration build
+    has completed successfully.
+    This way, the cache will only get filled up
+    with derived files that are part of a complete, successful build
+    not with files that might be later overwritten
+    while you debug integration problems.
+
+    </para>
+
+    <para>
+
+    In this case, you can use the
+    the <literal>--cache-force</literal> option
+    to tell &SCons; to put all derived files in the cache,
+    even if the files already exist in your local tree
+    from having been built by a previous invocation:
+
+    </para>
+
+    <screen>
+      % <userinput>scons -Q --cache-disable</userinput>
+      cc -o hello.o -c hello.c
+      cc -o hello hello.o
+      % <userinput>scons -Q -c</userinput>
+      Removed hello.o
+      Removed hello
+      % <userinput>scons -Q --cache-disable</userinput>
+      cc -o hello.o -c hello.c
+      cc -o hello hello.o
+      % <userinput>scons -Q --cache-force</userinput>
+      scons: `.' is up to date.
+      % <userinput>scons -Q</userinput>
+      scons: `.' is up to date.
+    </screen>
+
+    <para>
+
+    Notice how the above sample run
+    demonstrates that the <literal>--cache-disable</literal>
+    option avoids putting the built
+    <filename>hello.o</filename>
+    and
+    <filename>hello</filename> files in the cache,
+    but after using the <literal>--cache-force</literal> option,
+    the files have been put in the cache
+    for the next invocation to retrieve.
+
+    </para>
+
+  </section>
+
+  <section>
+  <title>Minimizing Cache Contention:  the <literal>--random</literal> Option</title>
+
+    <para>
+
+    If you allow multiple builds to update the
+    shared cache directory simultaneously,
+    two builds that occur at the same time
+    can sometimes start "racing"
+    with one another to build the same files
+    in the same order.
+    If, for example,
+    you are linking multiple files into an executable program:
+
+    </para>
+
+    <programlisting>
+       Program('prog',
+               ['f1.c', 'f2.c', 'f3.c', 'f4.c', 'f5.c'])
+    </programlisting>
+
+    <para>
+
+    &SCons; will normally build the input object files
+    on which the program depends in their normal, sorted order:
+
+    </para>
+
+    <screen>
+      % <userinput>scons -Q</userinput>
+      cc -o f1.o -c f1.c
+      cc -o f2.o -c f2.c
+      cc -o f3.o -c f3.c
+      cc -o f4.o -c f4.c
+      cc -o f5.o -c f5.c
+      cc -o prog f1.o f2.o f3.o f4.o f5.o
+    </screen>
+
+    <para>
+
+    But if two such builds take place simultaneously,
+    they may each look in the cache at nearly the same
+    time and both decide that <filename>f1.o</filename>
+    must be rebuilt and pushed into the shared cache directory,
+    then both decide that <filename>f2.o</filename>
+    must be rebuilt (and pushed into the shared cache directory),
+    then both decide that <filename>f3.o</filename>
+    must be rebuilt...
+    This won't cause any actual build problems--both
+    builds will succeed,
+    generate correct output files,
+    and populate the cache--but
+    it does represent wasted effort.
+
+    </para>
+
+    <para>
+
+    To alleviate such contention for the cache,
+    you can use the <literal>--random</literal> command-line option
+    to tell &SCons; to build dependencies
+    in a random order:
+
+    </para>
+
+    <!--
+
+    The following <screen> output was generated by this:
+
+    <scons_output example="ex-random">
+      <scons_output_command>scons -Q - -random</scons_output_command>
+    </scons_output>
+
+    We captured it directly here to guarantee a "random" order,
+    guarding against the potential for - -random to happen
+    to return things in the original sorted order.
+
+    -->
+
+    <screen>
+      % <userinput>scons -Q --random</userinput>
+      cc -o f3.o -c f3.c
+      cc -o f1.o -c f1.c
+      cc -o f5.o -c f5.c
+      cc -o f2.o -c f2.c
+      cc -o f4.o -c f4.c
+      cc -o prog f1.o f2.o f3.o f4.o f5.o
+    </screen>
+
+    <para>
+
+    Multiple builds using the <literal>--random</literal> option
+    will usually build their dependencies in different,
+    random orders,
+    which minimizes the chances for a lot of
+    contention for same-named files
+    in the shared cache directory.
+    Multiple simultaneous builds might still race to try to build
+    the same target file on occasion,
+    but long sequences of inefficient contention
+    should be rare.
+
+    </para>
+
+    <para>
+
+    Note, of course,
+    the <literal>--random</literal> option
+    will cause the output that &SCons; prints
+    to be inconsistent from invocation to invocation,
+    which may be an issue when
+    trying to compare output from different build runs.
+
+    </para>
+
+    <para>
+
+    If you want to make sure dependencies will be built
+    in a random order without having to specify
+    the <literal>--random</literal> on very command line,
+    you can use the &SetOption; function to
+    set the <literal>random</literal> option
+    within any &SConscript; file:
+
+    </para>
+
+    <programlisting>
+       Program('prog',
+               ['f1.c', 'f2.c', 'f3.c', 'f4.c', 'f5.c'])
+
+       SetOption('random', 1)
+       Program('prog',
+               ['f1.c', 'f2.c', 'f3.c', 'f4.c', 'f5.c'])
+    </programlisting>
+
+  </section>
+
+  <!--
+
+  <section>
+  <title>Troubleshooting Shared Caching:  the &cache-debug; Option</title>
+
+    <para>
+
+    XXX describe the - - cache-debug option
+    XXX maybe point to the troubleshooting appendix?
+
+    </para>
+
+  </section>
+
+  -->
+
+  <!--
+
+  <section>
+
+    <para>
+
+    XXX describe CacheDir management:  monitoring, deleting, etc.
+
+    </para>
+
+  </section>
+
+  -->