~and471/harvest/pretty-harvest

        <link rel="stylesheet" type="text/css" href="../theme/css/style.css" />

        <title>HACKING Harvest</title>

<div id="float">

<div id="topcap"></div>

<div id="round">

        <div id="header">

                <div id="logo-floater">

              <a href="http://daniel.holba.ch/harvest/" title="Harvest">

               <img alt="Ubuntu" id="logo" src="../theme/images/harvest.png" />

              </a>

        </div>

        <form action="handler.py" id="pkg-search-box">

            <div>

              <input type="text" name="pkg" size="27" />

              <input type="submit" value="Search Package Name" />

            </div>

         </form>

         <div id="sitename">

            <a href="http://daniel.holba.ch/harvest/">

              <img alt="" src="https://help.ubuntu.com/htdocs/ubuntunew/img/help-about.png" />

              <span>Finding low-hanging fruit...</span>

            </a>

          </div>

     </div>

 

        <div id="page">

                <div id="content">

                        <h1>HACKING Harvest</h1>

                        <p>The code of harvest lives in <a href="https://code.launchpad.net/harvest">Launchpad</a> and makes use of <a href="http://python.org/">python</a> and <a href="https://storm.canonical.com/">Storm</a>.</p>

                        <p>Harvest regularly pulls data from URLs stored in <a href="https://code.launchpad.net/harvest-data">this branch</a>. The file layout is pretty simple:</p>

                        <pre>daniel@bert:~/bzr/harvest-data$ ls

                        <p>Before attempting to download the CSV (comma-separated values) file, Harvest will check the <tt>Last-Modified</tt> entry in the HTTP header and see if any modifications were made. This is done to reduce traffic.</p>

                        <h2>Opportunities?</h2>

                        <p>The <tt>opportunities</tt> file is in CSV and of the following format:</p>

                        <pre>&lt;url&gt;,&lt;description&gt;</pre>

                        <p>The URLs to CSV files must be reachable via HTTP(s). The <tt>description</tt> is optional.</p>

                        <p>The CSV file in turn needs to be of the following form:</p>

                        <pre>&lt;sourcepackage&gt;,&lt;url&gt;,&lt;description&gt;</pre>

                        <p>For example:</p>

                        <pre>vdrift,http://launchpad.net/bugs/106854,106854</pre>

                        <p>Opportunities can be anything:</p>

                        <ul>

                                <li>Easy bugs</li>

                                <li>Suitable Upstream patches</li>

                                <li>Patches of other distributions</li>

                                <li>Problems in the CD builds that should be fixed</li>

                                <li>etc.</li>

                        </ul>

                        <p>Let your imagination go wild. :-)</p>

                        <h2>Clues?</h2>

                        <p>The <tt>clues</tt> file is in CSV and of the following format:</p>

                        <pre>&lt;url&gt;,&lt;score&gt;,&lt;description&gt;</pre>

                        <p>The URL specifies the link to another CSV file that should be pulled regularly.

                        The score is a float value that describes how good or bad it is for the package to be on the list

                        (eg. if a package is uninstallable that might be worth a -500, if 50% of the bugs are forwarded upstream that might be worth +300).

                        The scores are summed up every time the HTML pages are generated and might indicate if the package is in a good shape.</p>

                        <p>The format of the CSV file containing the <tt>clues</tt> is the same as that of the opportunities, right now only the source package name is used.</p>

                        <h2>Setting it up</h2>

                        <ol>

                                <li><tt>sudo apt-get install libapache2-mod-python python-storm bzr</tt></li>

                                <li><tt>mkdir ~/bzr; cd ~/bzr; bzr init-repo harvest; cd harvest; bzr branch lp:harvest</tt></li>

                                <li><tt>cd /var/www; sudo ln -s ~/bzr/harvest</tt></li>

                                <li>Edit one of the files in <tt>/etc/apache2/sites-enabled/</tt> to have contain a section like this one

                                        <pre>

&lt;Directory /var/www/harvest/&gt;

        AllowOverride FileInfo

        Order allow,deny

        allow from all

 &lt;/Directory&gt;</pre>

                                        to make sure <tt>.htaccess</tt> in the harvest tree is used.

                                </li>

                                <li><tt>sudo /etc/init.d/apache2 restart</tt></li>

                        </ol>

                <div id="footer">

            <div id="ubuntulinks">

                                <p class="copyright">&copy; 2008 Canonical Ltd.</p>

                        </div>

                </div>

 

                </div>

                </div>

                </div>

                <div id="bottomcap"></div>

</div>