diff --git a/pim_search.php b/pim_search.php
new file mode 100644
index 0000000..dad42a0
--- /dev/null
+++ b/pim_search.php
@@ -0,0 +1,147 @@
+<html>
+<div style="{width:400px; position:absolute; right:0px; background-color:#F0F8FF; margin:10px; border:2px solid #ddf; padding:5px;}">
+<h2>To do</h2>
+<ul>
+	<li>List the non working queries and debug them (use = keyword_to_test for comparison)</li>
+	<ul>
+		<li><a href="/search/full text search">full text search</a> should return <a href="http://fabien.benetou.fr/Events/RailsCampParis3#FullTextSearch">#FullTextSearch</a></li>
+		<li><a href="/search/facebook">facebook</a></li>
+		<li><a href="/search/google">google</a></li>
+		<li><a href="/search/apple">apple</a></li>
+		<li><a href="/search/Brooks">Brooks</a></li>
+		<li>... (cf log analysis "?bug&amp;keyword=problematickeyword")</li>
+	</ul>
+	<li>see also <a href="http://fabien.benetou.fr/Wiki/LearningSearch#ProblematicQueries">ProblematicQueries</a></li>
+	<li>make it more generalist via e.g. <a href="http://www.onelook.com/reverse-dictionary.shtml">reverse dictionary</a></li>
+	<li>better split documents, e.g. logs per day</li>
+	<li>handle the histories (wiki and repository diffs)</li>
+	<li>make URLs work, consider specific stemming</li>
+	<li>improving weightning and thus ranking</li>
+	<li>properly handle remote URLs</li>
+	<li>leverage .pageindex of each wiki</li>
+	<li>work on snippets</li>
+	<li>replace WikiName: by their icon, cf localmap</li>
+	<li>add pagerank and other topology metrics as attributes</li>
+	<li>use delta or RT indexing, requires better ID generation first though</li>
+	<li>add visualization, e.g. network or linked document with the same color</li>
+	<li>provide links to specific interesting queries, e.g. only this week</li>
+	<ul>
+		<li><a href="/search/%22internet%20brain%22~10">"internet brain"~10</a></li>
+	</ul>
+	<li><a href="http://fabien.benetou.fr/Wiki/LearningSearch">LearningSearch</a></li>
+	<li><a href="http://fabien.benetou.fr/Tools/Sphinxsearch">update Sphinxsearch</a></li>
+	<li>integrate <a href="http://fabien.benetou.fr/MemoryRecalls/ImprovingPIM#SocialPIM">Social PIM</a>, especially if there are 0 results</li>
+	<li>integrate external search, e.g. Seeks, especially if there are 0 results including through Social PIM</li>
+	</ul>
+</ul>
+</div>
+<div style="{width:500px; position:absolute; left:0px; background-color:#F8F0FF; margin:10px; border:2px solid #ddf; padding:5px;}">
+<?php
+
+# Overall this is very specialized and should be configurable to be used on other PIMs
+
+# this should be tested first and if not present, explain in a line where to download and how to install Sphinx API
+require ( "sphinxapi.php" );
+require ( "sphinx_doc_ids.php" );
+
+$index = "pmwikis";
+$words = $_GET["query"];
+if ( isset ($_GET["startingitem"]) )
+	$startingitem = $_GET["startingitem"];
+else 
+	$startingitem = 0;
+
+$image_path = "/devpim/pub/";
+$copy_img = "<img src=\"".$image_path."clipboard_add.png"."\" />";
+$star_img = "<img src=\"".$image_path."yellow-star.gif"."\" />";
+$alt_img = "<img src=\"".$image_path."server.png"."\" />";
+$bug_img = "<img src=\"".$image_path."bug.gif"."\" />";
+
+if ($words == ""){
+	print "You need to make an actual search, use ?query=keyword<br/>";
+	print "<form action=\"/devpim/pub/search.php\" method=\"get\"><input type=\"text\" name=\"query\"><input type=\"submit\" value=\"search\"/></form><br/>";
+	return;
+}
+if ($startingitem  == "" || $startingitem < 0){
+	$startingitem=0;
+}
+
+$itemspan=20;
+
+$cl = new SphinxClient ();
+$cl->SetLimits($startingitem, $startingitem+$itemspan);
+$cl->SetMatchMode(SPH_MATCH_EXTENDED2);
+$cl->SetRankingMode(SPH_RANK_PROXIMITY_BM25);
+# consider other ranking e.g. pagerank, weighted with freshness and hybrid
+$cl->SetSortMode(SPH_SORT_RELEVANCE);
+$res = $cl->Query($words,$index);
+#cf http://sphinxsearch.com/docs/manual-2.0.1.html#api-funcgroup-querying
+#$res = $cl->BuildExcerpts ( $docs, $index, $words, $opts );
+if ( !$res ) {
+	die ( "ERROR: " . $cl->GetLastError() . ".\n" );
+} else {
+	// XXX should test for empty result before doing the assigment
+	// if (0 results) { $socialsearch = shell_exec("pmwiki_social_search "$keyword"); }
+	// ideally this would be done ascynhroneously as it does take some time (few seconds!)
+	// yet still always offer it as an option (since it's "costly")
+	#var_dump($res);
+	$IDs = array_keys($res["matches"]);
+	print "<h2>Query<a href=\"?bug&keyword=$words\">$bug_img</a></h2>";
+
+	print "\"$words\" found ".$res["words"]["$words"]["hits"]." time(s) in ".$res["total_found"]." document(s):<br/>";
+	print "<ul>";
+	foreach ($IDs as $i)
+	{
+		#BuildExcerpts()
+		#consider here adding snippet, would require few disk access, MySQL storage could be faster...
+		print "<li>";
+		$file=$idx["$i"];
+		if (preg_match('|/home/fabien/irclogs/|',$file)){
+			$target = preg_replace('|/home/fabien/irclogs/(.*)|','$1',$file);
+			$url = urlencode($target);
+			print "<a href=\"https://cloud.benetou.fr/discussions/$url\">IRClogs:$target</a>";
+		}
+		elseif (preg_match('|/home/fabien/www/mirrors/|',$file)){
+			$page = preg_replace('|/home/fabien/www/mirrors/(.*)/wiki.d/(.*)\.(.*)|','$2/$3',$file);
+			$source = preg_replace('|/home/fabien/www/mirrors/(.*)/wiki.d/(.*)\.(.*)|','$1',$file);
+			switch ( $source ) {
+				case "saint-maur": $url="http://saint-maur.benetou.fr"; break;
+				case "fabien": $url="http://fabien.benetou.fr"; break;
+				case "pim": $url="http://www.ourp.im/"; break;
+				case "agiwiki": $url="http://www.agi-wiki.org/"; break;
+				case "wiki": $url="http://fabien.benetou.fr/innovativ.it/www/HistoricalArchives/Seedea"; break;
+				default: $url=$source;
+			}
+			print "<a href=\"$url/$page\">$source:$page</a>";
+			print preg_replace('|/home/fabien/www/mirrors/(.*)/wiki.d/(.*)\.(.*)|','<a href="http://self/mirrors/$1/$2/$3">'.$alt_img.'</a>',$file);
+			print "<a href=\"javascript:[[$page]]\">$copy_img</a>";
+		} elseif (preg_match('|/home/fabien/repository/|',$file)){
+			print preg_replace('|/home/fabien/repository/(.*)|','<a href="http://fabien.benetou.fr/repository/?p=.git;a=blob;f=$1">Repository:$1</a>',$file);
+		} else {
+			print preg_replace('|/home/fabien/www/(.*)/wiki.d/(.*)\.(.*)|','<a href="http://self/$1/$2/$3">$1:$2/$3</a>',$file);
+			print preg_replace('|/home/fabien/www/(.*)/wiki.d/(.*)\.(.*)|','<a href="https://cloud.benetou.fr/backups/wiki/$2/$3">'.$alt_img.'</a>',$file);
+		}
+		print "<a href=\"?star&keyword=$words&result=$file\">$star_img</a>";
+		print "</li>";
+	}
+	$previousitems=$startingitem-$itemspan;
+	$nextitems=$startingitem+$itemspan;
+	print "</ul>";
+	print "<div>";
+	if ($previousitems >= 0)
+		print "<a href=\"/search/$words/$previousitems\">previous items</a>";
+	for ($p=1;$p<$res["total"]/$itemspan;$p++)
+		print " <a href=\"/search/$words/".($p-1)*$itemspan."\">$p</a>/";
+	if ($nextitems < $res["total"])
+		print "<a href=\"/search/$words/$nextitems\">next items</a>";
+	print "</div>";
+
+	$indexage = "unknown";
+	$targetindexfile = "/var/lib/sphinxsearch/data/pmwikis.spd";
+	$updatedindex = filemtime($targetindexfile);
+	$indexage = date("c",$updatedindex);
+	print "<hr/><center>$indexage index<br/>(if bug check indexer errors).</center>";
+}
+
+?>
+</div>
