LHC@home

The latest version of the Wiki software is out. I am pretty sure I know what to do to implement a "Goggle" search of the Wiki, which will, at least initially be just an enhancement to the search box to use the Google search engine.

The intent is to change from the baseline search which uses the MySQL "fulltext-search" which is a word only search. So, it finds occurrences of words, and does not allow searching for phrases etc.

So, health permitting, Monday I will:

1) Take BOINC Wiki off-line if needed.
2) Make a database back up
2a) download image directories
3) Install media Wiki 1.5.0
3a) Upload image directories
4) Make selected changes to implement Google Search in the current search dialog box, thus allowing more concise searches.
5) Not sure ...
6) Bring the system back alive.

With luck, no mistakes, who knows ...

---

My test version is up now on the site...

Gonna take a break ... it is 9 PM California (PST) time ... and, I have most of the first steps done. I proofed I can duplicate the site, made the code changes and tested them to see that the work.

Problem is, I have not tested the conversion of the site to the new version and made the code changes there ...

THEN, I have to truncate some data, make sure that does not break the data base, make sure the internal links are properly updated ...

THEN, I can start on this for real ... gee... I only seem to have 36 hours into this project ... :)

The good news is that I can turn on both searches, only one or the other with a configuration parameter change. And, yes, I plan to submit the code once I see that it works with the new version ...

Key feature is that you can use the new or the old familiar search. In the Google search you can use double quotes now to look for exact phrases.

So, enter Quorum of Results in both searches, look at the words highlighted. Then, do "Quorum of Results" and see the difference ... in essence it should cut down on the number of spurious pages pulled.

And, yes, I had planned to look into an advanced searching page ...

The depressing thing was how much PHP code I had forgotten ... :(

# The type, or types of searches allowed. Values are: "Google", "WikiDefault", "Both"
$wgSearchTypeEnabled = "Both";

and:

    <?php
    global $wgSearchTypeEnabled; // pull in the externally declared "Global"
    if ((strtolower($wgSearchTypeEnabled) == strtolower("Google")) ||
        (strtolower($wgSearchTypeEnabled) == strtolower("Both"  ))) {
      echo "    <div id="p-search" class="portlet">\r\n";
      echo "      <h5><label for="searchInput">Google Search</label></h5>\r\n";
      echo "      <div class="pBody">\r\n";
      echo "        <!-- SiteSearch Google -->\r\n";
      echo "        <form method="get" \r\n";
      echo "              action="http://www.google.com/custom" \r\n";
      echo "              target="_top">\r\n";
      echo "          <table border="0" bgcolor="#ffffff">\r\n";
      echo "          <tr>\r\n";
      echo "              <td nowrap="nowrap" valign="top" align="center" height="32">\r\n";
      echo "                <a href="http://www.google.com/" title="Google Logo">\r\n";
      echo "                  <img src="http://www.google.com/logos/Logo_25wht.gif"\r\n";
      echo "                    border="0" \r\n";
      echo "                       alt="Google">\r\n";
      echo "                  </img>\r\n";
      echo "                </a><br/>\r\n";
      echo "                 <input type="hidden" \r\n";
      echo "                        name="domains"\r\n";
      echo "                       value="http://boinc-doc.net/boinc-wiki/" />\r\n";
      echo "                 <input type="hidden" \r\n";
      echo "                        name="sitesearch" \r\n";
      echo "                       value="http://boinc-doc.net/boinc-wiki/" />\r\n";
      echo "                 <input   id="searchInput"\r\n";
      echo "                        type="text" \r\n";
      echo "                        name="q" \r\n";
      echo "                       value=""></input>\r\n";
      echo "              </td>\r\n";
      echo "            </tr>\r\n";
      echo "            <tr>\r\n";
      echo "              <td nowrap="nowrap" align="center">\r\n";
      echo "                <input type="submit" name="sa" value="Google Search" class="searchButton" />\r\n";
      echo "                <input type="hidden" name="ie" value="UTF-8"></input>\r\n";
      echo "                <input type="hidden" name="oe" value="UTF-8"></input>\r\n";
      echo "                <input type="hidden" name="hl" value="en"></input>\r\n";
      echo "              </td>\r\n";
      echo "            </tr>\r\n";
      echo "          </table>\r\n";
      echo "        </form>\r\n";
      echo "        <!-- SiteSearch Google --> \r\n";
      echo "      </div>\r\n";
      echo "    </div>\r\n";
    } 
    
    if ((strtolower($wgSearchTypeEnabled) == strtolower("WikiDefault")) ||
        (strtolower($wgSearchTypeEnabled) == strtolower("Both"  ))     ) {
      echo "    <div id="p-search" class="portlet">\r\n";
      echo "      <h5><label for="searchInput">";
      $this->msg('search');
      echo "</label></h5>\r\n";
      echo "      <div class="pBody">\r\n";
      echo "        <!-- SiteSearch WikiDefault --> \r\n";
      echo "        <form    name="searchform" \r\n";
      echo "               action="";
      $this->text('searchaction');
      echo ""\r\n";
      echo "                   id="searchform">\r\n";
      echo "          <input   id="searchInput" \r\n";
      echo "                 name="search"\r\n";
      echo "                 type="text"\r\n";
      if ($this->haveMsg('accesskey-search')) {
        echo "            accesskey="";
        $this->msg('accesskey-search'); 
        echo ""\r\n";
      }
      if (isset($this->data['search']) ) {
        echo "                value="";
        $this->text('search');
        echo """; // don't end with line feed, need end of control
      }
      echo " />\r\n";
      echo "          <input type="submit"\r\n";
      echo "                 name="go"\r\n";
      echo "                class="searchButton"\r\n";
      echo "                   id="searchGoButton"\r\n";
      echo "                value="";
      $this->msg('go');
      echo "" /> \r\n";
      echo "          <input type="submit"\r\n";
      echo "                 name="fulltext"\r\n";
      echo "                class="searchButton"\r\n";
      echo "                value="";
      $this->msg('Search');
      echo "" />\r\n";
      echo "        </form>\r\n";
      echo "        <!-- SiteSearch WikiDefault --> \r\n";
      echo "	  </div>\r\n";
      echo "	</div>\r\n";
    
    }

    ?>

Most of the original code from the two examples, the one in the baseline page, and the one person who suggested a possible solution are basically there, just neatened up so it is readable and maintainable ...

---

As you know Paul,our computers hook us and see if they can keep us awake 24/7

I tend to do this data processing around the clock (my dialup makes me)and have been messing with websites for several years too.




<B>*____* [/b]

Hey look what I found using google on the BOINC Wiki! )





Nice work Paul

---

Volunteer Mad Scientist For Life

Well, after two, or is it three LONG days, I think I know what I have to do ...

The good news is that it looks like it should be straight forward now. I just have to do the remote stuff. I did a complete download of the database, installed it locally, made it work, then did installations of the latest version (I really like what they have done with it ...).

Since 4 something last night I have been plugging away at the code, and you see what I did basically. From there, it was working on the changed configuration and in THAT process I made another small code change to make it more consistant (most people cannot edit, so they should not see an "Upload" link).

Now, I have a full set of installable files, I will make a couple more small changes, upload those, then take the system off line and do another database back up (which I have to do manally for each of the tables).

The good news, if you will, is that I am planning to drop all of the edit history to save database space (the 1.5 grew my space from 48M to 65M, almost all of which is history we mostly don't care about).

So, that will save me the biggest back up problem to start.

The actual "migration" should only take a few minutes. So, the biggest time waste now is doing the rest of the file fiddling (I am also cleaning up lots of old junk floating around the system.

---

Ok, I am taking the site off-line and it will be back, I hope, in a couple of hours ... I am talking programmer time, so, it could be till next Tuesday ...

I will post an update here, and in a new thread when it is back up ...

---

Ok, it seems to be working for me ...

Remember, you do not need an account to use the Wiki.

Just look around...

The two search boxes do different searches. The lower one works as before. The "Google" one, works as the BOINC Web Sites do now, it searches the BOINC Wiki and will pull up a "better" list. You can do the usual search operations that were not available before.

The simplest is the one I described below. You can also try anther like BOINC System ... with just the two words you get a result like:

Main Page - BOINCWiki
The material that comprises the BOINC System in all of its glory can be found in
the following topical categories:. Participating in BOINC Powered Projects ...

With the "hit" count at 867 ...

using "BOINC System" ... the second word is no longer highlighted. The "hit" count is 283 ...

---

There seems to be some lingering "oddness". Some of the links seem to want to report themselves in:

http://boinc-doc.net/b-w/index.php?title=

Instead of the correct:

http://boinc-doc.net/boinc-wiki/index.php?title=

I cannot find where this "magic" is happening TO me ... but, if I can find it I will see what I can do to eradicate it.

In the mean time, SOMETIMES, manually editing the link to the correct link seems to fix the problem, and MAY solve it for all of us. I have not proved this "fix" yet. Editors, you can "force" an "empty" change which also seems to cure the link error. You can tell if the link is bad floating the cursor over the link.

Report any link errors to me, e-mail is "best" (ugh, the work!), as well as any other problems. For all of that, I really do think that this is a HUGE improvement. We have many new features, including:

# related changes,
# printable version (a requested feature, but you still have to visit all pages you want to print
# New page list
# "Special Pages" page list is alphabetized
# Google search (Yea! Paul! :))
# Changes to the security model (makes control easier, and removes some things the normal viewer cannot use, like the IP Link at the top of the page ... Oh, and I changed the wording of the message from "Create account ..." to only "Log In", still going to get calls though
# FDL Icon, and stuff ...

Add other features as you find them ...

---

Yes it does seem to be Magical when you are doing this!


(you probably noticed that websites will ask you to use that arrow thing instead of the [] )

LOL.....you can't even type that pointy thing without it messing up your post here

---

Volunteer Mad Scientist For Life

Well, LHC@HOme and Predictor@Home I think are the only ones that don't use the BBCode version of the web site.

So, when i pasted I did not bother to fix it. when they do update the sofware the message will be corrcet fr future generations.

---

Yeah I noticed that here but I tend to run into that on a few other non-scientific sites and others like yahoo that went with Seti Classic groups with nothing at all[/b] (which I don't like at all)

But I am a guy that can't stand his own typo's so I tend to check and edit any of my screw ups no matter how many time it takes


I do prefer BBCode itself rather than HTML.......just typing it out is faster and easier to me.

---

Volunteer Mad Scientist For Life

Yes, well, I strive for accuracy. But, this time I was just too tired to care. I have only been able to sleep about 4-5 hours every 36 to 48 ... so, I have periods when I am just not able do much, and that was one of them. Not sorry, I do like the way the site came out so it was worth the "pain".

Amd we are improving ... as the google search settles in, I can look to the next features to add. We *DO* want to add more value, I just have no clear idea of what I can do to make things better.

Some of the suggestions founder on technical capabities and "expense" of development vs. possible utility. Anyways ...

---

You do a very nice job Paul.

I am the same with my sleep vs time on the computers.

(as you can tell since this is 3:30am PT)


Most of my life I have not been able to attempt sleep until about 4am and for the last 4yrs I tend to be awake for those 36 to 48hr stretches.


The nice thing is now that we have these machines it gives us something of *value* to do





Keep up the good work Paul......and take care of yourself too.

---

Volunteer Mad Scientist For Life