tutorial-debug.xml

Delivered as text/xml

[ hide source ] | [ make this the default ]

File Contents

<?xml version='1.0' ?>
<!DOCTYPE sect1 PUBLIC "-//OASIS//DTD DocBook XML V4.4//EN"
               "http://www.oasis-open.org/docbook/xml/4.4/docbookx.dtd" [
<!ENTITY % myvars SYSTEM "../variables.ent">
%myvars;
]>
<sect1 id="tutorial-debug">
  <title>Debugging and Automated Testing</title>
  <authorblurb>
    <para>by <ulink url="mailto:joel@aufrecht.org">Joel Aufrecht</ulink></para>
  </authorblurb>
  
  <sect2><title>Debugging</title>
    <formalpara>
      <title>Developer Support</title>
      <para>The Developer Support package adds several goodies: debug
      information for every page; the ability to log comments to the
      page instead of the error log, and fast user switching so that you
      can test pages as anonymous and as dummy users without logging
      in and out.</para>
    </formalpara>
    <formalpara>
      <title>PostgreSQL</title>
      <para>You can work directly with the database to do debugging
          steps like looking directly at tables and testing stored
          procedures.  Start emacs.  Type
            <userinput>M-x sql-postgres</userinput>.  Press enter for
            server name and use <userinput><replaceable>$OPENACS_SERVICE_NAME</replaceable></userinput> for
            database name.  You can use C-(up arrow) and C-(down arrow)
            for command history.
</para>
    </formalpara>
    <para>Hint: "Parse error near *" usually means that an xql file
      wasn&#39;t recognized, because the Tcl file is choking on the *SQL*
      placeholder that it falls back on.</para>
    <formalpara>
      <title>Watching the server log</title>
      <para></para>
    </formalpara>
    <para>To set up real-time monitoring of the AOLserver error
          log, <emphasis role="bold">type</emphasis> <screen>less /var/lib/aolserver/<replaceable>$OPENACS_SERVICE_NAME</replaceable>/log/openacs-dev-error.log</screen>
          <literallayout>F to show new log entries in real time (like tail -f)
C-c to stop and F to start it up again. 
G goes to the end.
? searches backward 
/ searches forward. 
          </literallayout>
    </para>
  </sect2>
  <sect2>
    <title>Manual testing</title>
        <para>Make a list of basic tests to make sure it works</para>
        <segmentedlist>
          <?dbhtml list-presentation="table"?>
          <segtitle>Test Num</segtitle>
          <segtitle>Action</segtitle>
          <segtitle>Expected Result</segtitle>
          <seglistitem>
            <seg>001</seg>
            <seg>Browse to the index page while not logged in and
            while one or more notes exist.</seg>
            <seg>No edit or delete or add links should appear.</seg>
          </seglistitem>
          <seglistitem>
            <seg>002</seg>
            <seg>Browse to the index page while logged in.  An Edit
            link should appear.  Click on it.  Fill out the form and
            click Submit.</seg>
            <seg>The text added in the form should be visible on the
            index page.</seg>
          </seglistitem>
          <seglistitem>
            <seg>API-001</seg>
            <seg>Invoke mfp::note::create with a specific word as the title.</seg>
            <seg>Proc should return an object id.</seg>
          </seglistitem>
          <seglistitem>
            <seg>API-002</seg>
            <seg>Given an object id from API-001, invoke mfp::note::get.</seg>
            <seg>Proc should return the specific word in the title.</seg>
          </seglistitem>
          <seglistitem>
            <seg>API-003</seg>
            <seg>Given the object id from API-001, invoke mfp::note::delete.</seg>
            <seg>Proc should return 0 for success.</seg>
          </seglistitem>
        </segmentedlist>
        <para>Other things to test: try to delete someone else&#39;s
        note.  Try to delete your own note.  Edit your own note.
        Search for a note.</para>
  </sect2>
  
  <sect2>
    <title>Write automated tests</title>

    <authorblurb>
      <para>by <ulink url="mailto:simon@collaboraid.net">Simon Carstensen</ulink> and Joel Aufrecht</para>
    </authorblurb>

    <para><indexterm><primary>Automated tests</primary></indexterm>
    It seems to me that a lot of people have been asking for some guidelines on how to write automated tests. I&#39;ve done several tests by now and have found the process to be extremely easy and useful. It&#39;s a joy to work with automated testing once you get the hang of it.</para>
    <para>Create the directory that will contain the test
    script and edit the script file.  The directory location and filename are standards which are recognized by the automated testing package:</para>
    <screen>[$OPENACS_SERVICE_NAME www]$<userinput> mkdir /var/lib/aolserver/<replaceable>$OPENACS_SERVICE_NAME</replaceable>/packages/myfirstpackage/tcl/test</userinput>
[$OPENACS_SERVICE_NAME www]$<userinput> cd /var/lib/aolserver/<replaceable>$OPENACS_SERVICE_NAME</replaceable>/packages/myfirstpackage/tcl/test</userinput>
[$OPENACS_SERVICE_NAME test]$ <userinput>emacs myfirstpackages-procs.tcl</userinput></screen>
    <para>Write the tests.  This is obviously the big step :)  The script should first call ad_library like any normal -procs.tcl file:</para>
    <screen>ad_library {
    ...
}
</screen>
        <para>To create a test case you call
<computeroutput><ulink url="/api-doc/proc-view?proc=aa%5fregister%5fcase">aa_register_case</ulink> test_case_name.</computeroutput>.   
Once you&#39;ve created the test case you start writing the needed logic.
We&#39;ll use the tutorial package, "myfirstpackage," as an example.
Let&#39;s say you just wrote an <ulink url="/api-doc">API</ulink> for adding and deleting notes in the
notes packages and wanted to test that. You&#39;d probably want to write a
test that first creates a note, then verifies that it was inserted,
then perhaps deletes it again, and finally verifies that it is
gone.</para> 

<para>
Naturally this means you&#39;ll be adding a lot of bogus data to the
database, which you&#39;re not really interested in having there. To avoid
this I usually do two things. I always put all my test code inside a
call to aa_run_with_teardown which basically means that all the
inserts, deletes, and updates will be rolled back once the test has
been executed. A very useful feature. Instead of inserting bogus data
like:        <computeroutput>set name "Simon"</computeroutput>, I tend to generate a random script in order avoid inserting a value that&#39;s already in the database:</para>
<screen>set name [ad_generate_random_string]
</screen>
<para>Here&#39;s how the test case looks so far:</para>

<screen>aa_register_case mfp_basic_test {
    My test
} {
    aa_run_with_teardown \
       -rollback \
       -test_code  {

       }
}
</screen>
<para>Now let&#39;s look at the actual test code. That&#39;s the code that
goes inside <computeroutput>-test_code {}</computeroutput>.  We want to implement test case API-001, "Given an object id from API-001, invoke mfp::note::get.  Proc should return the specific word in the title."</para>
    <programlisting>
      set name [ad_generate_random_string]
      set new_id [mfp::note::add -title $name]
      aa_true "Note add succeeded" {$new_id ne ""}</programlisting>
    <para>To test our simple case, we must load the test file into the system (just as with the /tcl file in the basic tutorial, since the file didn&#39;t exist when the system started, the system doesn&#39;t know about it.)  To make this file take effect, go to the <ulink url="/acs-admin/apm">APM</ulink> and choose "Reload changed" for "MyFirstPackage".  Since we&#39;ll be changing it frequently, select "watch this file" on the next page.  This will cause the system to check this file every time any page is requested, which is bad for production systems but convenient for developing.  We can also add some aa_register_case flags to make it easier to run the test.  The <computeroutput>-procs</computeroutput> flag, which indicates which procs are tested by this test case, makes it easier to find procs in your package that aren&#39;t tested at all.  The <computeroutput>-cats</computeroutput> flag, setting categories, makes it easier to control which tests to run.  The <computeroutput>smoke</computeroutput> test setting means that this is a basic test case that can and should be run any time you are doing any test. (<ulink url="http://www.nedbatchelder.com/blog/20030408T062805.html">a definition of "smoke test"</ulink>)</para>
      <para>Once the file is loaded, go to <ulink url="/test">ACS Automated Testing</ulink> and click on myfirstpackage.  You should see your test case.  Run it and examine the results.</para>
    <sect3>
      <title>TCLWebtest tests</title>
      <para>API testing can only test part of our package - it doesn&#39;t test the code in our adp/tcl pairs.  For this, we can use TCLwebtest.  TCLwebtest must be <link linkend="install-tclwebtest">installed</link> for this test to work.  This provides a <ulink url="http://tclwebtest.sourceforge.net/doc/api_public.html">library of functions</ulink> that make it easy to call a page through HTTP, examine the results, and drive forms.  TCLwebtest&#39;s functions overlap slightly with acs-automated-testing; see the example provided for one approach on integrating them.</para>
    </sect3>
    <sect3>
      <title>Example</title>
      <para>Now we can add the rest of the API tests, including a test with deliberately bad data.  The complete test looks like:</para>
      <programlisting><xi:include href="../../files/tutorial/myfirstpackage-procs.tcl" xi:parse="text" xmlns:xi="http://www.w3.org/2001/XInclude"><xi:fallback>example missing</xi:fallback></xi:include></programlisting>
      <para>See also <xref linkend="automated-testing-best-practices"/>.</para>
    </sect3>
  </sect2>
</sect1>