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, 353 insertions, 0 deletions

diff --git a/doc/python10/process.xml b/doc/python10/process.xml
new file mode 100644
index 0000000..f1b2479
--- /dev/null
+++ b/doc/python10/process.xml
@@ -0,0 +1,353 @@
+<para>
+
+  The &SCons; project has paid particular attention from day one to the
+  development process. One of the first internal documents produced was
+  a set of Developer's Guidelines to provide a loose framework for what
+  we were trying to accomplish and how we would go about accomplishing
+  it. These Guidelines cover things like:
+
+</para>
+
+<itemizedlist>
+
+  <listitem>
+    <para>
+
+      &SCons; will be written to Python version 1.5.2 (to ensure
+      usability by a wide install base).
+
+    </para>
+  </listitem>
+
+  <listitem>
+    <para>
+
+      How &SCons; is be tested: which infrastructure modules to use,
+      what platforms to test on, etc.
+
+    </para>
+  </listitem>
+
+  <listitem>
+    <para>
+
+      Expectations for developers (subscribe to the mailing list,
+      encouraged to register at SourceForge).
+
+    </para>
+  </listitem>
+
+  <listitem>
+    <para>
+
+      Brief outline of how to use the change management systems (Aegis and
+      CVS) for &SCons; development;.
+
+    </para>
+  </listitem>
+
+</itemizedlist>
+
+<para>
+
+  Establishing these guidelines up front had two purposes: 1)
+  Demonstrate the seriousness of the project to anyone wondering about
+  joining the effort; 2) Give potential developers an idea up front as
+  to whether their development style would mesh with the rest of the
+  project.
+
+</para>
+
+<section>
+  <title>Aegis</title>
+
+  <para>
+
+    One of the most important aspects of the &SCons; development process
+    is the use of Peter Miller's Aegis change management system. I
+    had been using Aegis for personal projects for several years, and
+    found its development methodology vastly improved the quality of my
+    programming. I was consequently committed to using it for &SCons;
+    development.
+
+  </para>
+
+  <para>
+
+    Aegis provides a number of things, including:
+
+  </para>
+
+  <itemizedlist>
+
+    <listitem>
+      <para>
+
+        A flexible source code control and branching model.
+
+      </para>
+    </listitem>
+
+    <listitem>
+      <para>
+
+        A defined process with separate development, review and
+        integration steps.
+
+      </para>
+    </listitem>
+
+    <listitem>
+      <para>
+
+        A distributed development model based on distribution of atomic
+        change sets.
+
+      </para>
+    </listitem>
+
+  </itemizedlist>
+
+  <para>
+
+    The single most important reason for using Aegis, however, is its
+    management of automated tests as part of the development process.
+
+  </para>
+
+</section>
+
+<section>
+  <title>Testing, Testing, Testing</title>
+
+  <para>
+
+    The &SCons; project has made extensive use of automated tests from day
+    one, taking inspiration mostly from Aegis, partly from the eXtreme
+    Programming model, and with a little home-brew scripting for glue.
+
+  </para>
+
+  <section>
+    <title>Testing Criteria</title>
+
+    <para>
+
+      The underlying criteria for testing changes to the &SCons; code
+      are taken from Aegis:
+
+    </para>
+
+    <itemizedlist>
+
+      <listitem>
+        <para>
+
+          Every change must have one or more new or modified tests
+          checked in along with the code.
+
+        </para>
+      </listitem>
+
+      <listitem>
+        <para>
+
+          The new code being checked in must pass all of the new and/or
+          modified tests.
+
+        </para>
+      </listitem>
+
+      <listitem>
+        <para>
+
+          The <emphasis>old</emphasis>, already checked-in code in must
+          <emphasis>fail</emphasis> all of the new and/or modified
+          tests.
+
+        </para>
+      </listitem>
+
+      <listitem>
+        <para>
+
+          The new code being checked in must pass all unmodified,
+          already checked-in tests.
+
+        </para>
+      </listitem>
+
+    </itemizedlist>
+
+    <para>
+
+      In practice, these restrictions can be overridden as necessary­for
+      example, when changing comments or documentation.
+
+    </para>
+
+    <para>
+
+      The criterion that surprises many people is having the old code
+      fail the tests in the change. This makes sure that the new tests
+      or modified tests really do exercise the bug fix or feature being
+      added by the change.
+
+    </para>
+
+    <para>
+
+      Together, these criteria ensure that every newly checked-in
+      version &SCons; conforms to defined behavior, as defined by
+      the tests. Whenever a bug is found, its fix is checked in with
+      a new or modified test that guarantees the bug will not recur
+      in the future. We have already built up a regression test base
+      of almost 90 tests that cover the vast majority of &SCons;'
+      functionality.
+
+    </para>
+
+  </section>
+
+  <section>
+    <title>Testing Infrastructure</title>
+
+    <para>
+
+      Testing standards are no good if they're too much of a burden for
+      developers, who will at best work around or ignore the testing
+      requirements, or at worst stop contributing code and go join a
+      project that's more fun. To this end, good testing infrastructure
+      that makes it easy to write tests is crucial.
+
+    </para>
+
+    <para>
+
+      &SCons; development uses two development methodologies, one for
+      the individual modules in the build engine, and the other for
+      end-to-end tests of the &SCons; script.
+
+    </para>
+
+    <para>
+
+      For the build engine modules, we use PyUnit. Every change to a
+      build engine module must have a change to its corresponding unit
+      tests, which live side-by-side in a separate file that imports
+      module. As we build up a large body of unit tests, this ensures
+      that the build engine will perform correctly whenever someone uses
+      it in some application other than the &SCons; script itself.
+
+    </para>
+
+    <para>
+
+      For end-to-end script tests, we have developed two modules to make
+      writing tests easy. The first, <filename>TestCmd.py</filename>,
+      is a generic module for
+      testing commands or scripts (in any language, not just Python).
+
+      The second module, <filename>TestScons.py</filename>,
+      is a subclass of the generic
+      <filename>TestCmd.py</filename> module.
+      <filename>TestScons.py</filename>
+      takes care of initialization and
+      displaying error conditions
+      specific to testing &SCons;.
+
+    </para>
+
+    <para>
+      
+      In practice, simple tests only
+      need to initialize a test object, use the object to write some
+      input files, run &SCons;, and then check whatever criteria
+      determine whether the test passed or failed. A complete test of
+      the &Program; method, for example, looks like this:
+
+    </para>
+
+    <programlisting>
+      test = TestSCons.TestSCons()
+
+      test.write('SConstruct',
+      """env = Environment()
+      env.Program(target = 'foo', source = 'foo.c')
+      """)
+
+      test.write('foo.c',
+      """
+      int
+      main(int argc, char *argv[])
+      {
+          argv[argc++] = "-"; /* dummy use of args */
+          printf("foo.c successfully compiled\\n");
+          exit (0);
+      }
+      """)
+
+      test.run(arguments = 'foo') # runs SCons
+
+      test.run(program = test.workpath('foo'))
+
+      test.fail_test(test.stdout() != "foo.c successfully compiled\n")
+
+      test.pass_test()
+    </programlisting>
+
+  </section>
+
+</section>
+
+<section>
+  <title>SourceForge</title>
+
+  <para>
+
+    Registration of the &SCons; project was approved at SourceForge on
+    29 June 2001.  Within a week, the initial code base was checked in,
+    mailing lists were created, and the web site was set up. We started
+    making use of the task-list manager to track what we had to finish
+    for initial release.
+
+  </para>
+
+  <para>
+
+    The obvious complication was how to use
+    structured testing methodology of Aegis when SourceForge uses
+    CVS for source control. Not using the SourceForge CVS tree would
+    have had two significant disadvantages: one, missing out on the
+    archiving and central location in the event of disaster; two, people
+    coming to the SourceForge project page wouldn't be able to browse
+    the source.  The latter was particularly important in
+    the early stages of development, in order to avoid any impression
+    that this was Yet Another Project that starts with a bang and then
+    dwindles as the initial enthusiasm starts to wear off.
+
+  </para>
+
+  <para>
+
+    The solution was to use the SourceForge CVS repository for read-only
+    access to the source. &SCons; developers are welcome to use CVS for
+    their development, but the changes are <emphasis>not</emphasis>
+    committed to the SourceForge repository. Instead, patches are sent
+    to the integrator for processing through Aegis. When the change
+    has been integrated into the Aegis repository, a home-brew
+    script translates the Aegis change into a virtual shell script
+    of commands that copy the necessary files from Aegis and check them
+    in to CVS at SourceForge.
+
+  </para>
+
+  <para>
+
+    (In practice, write access is not actually disabled for registered
+    developers, but if they do make any changes directly at SourceForge,
+    they can be overwritten at the next Aegis update.)
+
+  </para>
+
+</section>