So, I thought I would write this post as a reminder to myself of the process and to share my experience (and screw ups) with everyone. Here we go!

Upgrade

First, I went ahead and upgraded this blog to Umbraco 4.5.2. You can find an upgrade guide on our.umbraco.org.

MySql

To convert everything to MySql, I simply created a fresh Umbraco installation running MySql and installed a package (containing all of my content, templates, doc types, xslt, etc.) that I created from the original blog.

Here is where I screwed up.

To deploy my blog to Rackspace, I simply did a backup of my local MySql database and restored it to the server on Rackspace using MySql Administrator. I was left with an exception that told me that the table UMBRACODOMAINS did not exist. Of course it did exist and after some research I found a solution to the problem.

MySql has a configurable identifier called lower_case_table_names that has a different default value based on the OS that MySql resides on.

My local installation is on Windows XP, so the default value is 1. That means that all table names will be converted to lowercase and that name comparisons ARE NOT case sensitive.

That means when I installed Umbraco on my local server, all of my table names were converted to lowercase. However, queries from Umbraco use uppercase table names.  Again, this is not a probem on Windows becase comparisons are not case sensitive.

The problem is that Rackspace runs MySql on Linux where the default value is 0. That means that all table names are stored using the lettercase specified in the CREATE TABLE statement and comparisons ARE case sensitive.

Since I did a backup/restore from my local installation, all of the table names on the Rackspace server were in lowercase, thus causing the exception.

To solve the problem, I simply added the following under the [mysqld] section of C:\Program Files\MySQL\MySQL Server 5.1\my.ini:

[mysqld]
lower_case_table_names=0

Now, when I create local Umbraco installations running MySql, the table names will be left in uppercase making the databases interchangeable with MySql on Rackspace.

Web.config

Next, I set the following in web.config.

<add key="umbracoUseMediumTrust" value="true" />

Some instructions also tell you to uncomment the following line (as you would to force medium trust at the app level), but that will cause an exception on Rackspace because they do not allow app level overrides, so just leave it as is.

<!-- <trust level="Medium" originUrl=".*" />-->

Impersonation

Finally, you will need to add impersonation under the system.web section. This will enable Umbraco to read/write to the file system. You will need to impersonate the default FTP account that is created by Rackspace when you created the site.

If you are in the Dallas datacenter you will use the following:

<identity impersonate="true" userName="dfw\YOUR_FTP_USERNAME" password="YOUR_FTP_PASSWORD" />

If you are in the San Antonio datacenter you will use the following:

<identity impersonate="true" userName="sat\YOUR_FTP_USERNAME" password="YOUR_FTP_PASSWORD" />

Of course, you will also need to update umbracoDbDSN with the correct values for your database on Rackspace.

Drop File Extensions (optional)

Rackspace is already set up to run without file extensions, so you might as well go ahead and drop them.

<add key="umbracoUseDirectoryUrls" value="true" />

Be sure to update any rules you may have in /config/UrlRewriting.config as well.

Closing Thoughts

So far, my experience with Rackspace Cloud Sites has been great. Technical support (especially live chat) has been superb and the Control Panel is simple to use. I am incredibly grateful for the hard work and effort put in by Benjamin Howarth and the rest of the Umbraco team that enabled us to have Umbraco running in such an ideal environment.

However, after about three days on the interwebs, I decided otherwise. Bring on Akismet from Automattic. From their website, "Automattic Kismet (Akismet for short) is a collaborative effort to make comment and trackback spam a non-issue and restore innocence to blogging, so you never have to worry about spam again."

Sounds good to me, let's get started!

First, you will need a Wordpress account. You can then find your API key by clicking on Profile.

Next, you will need to download Akismet for .Net 2.0 from Codeplex and drop the included assembly into your bin directory. Now, time for some code.

I like the idea of saving every comment, but only publishing those that pass the spam check. So, my btnSubmit_Click event looks something like this (sorry for all the wraps):

    protected void btnSubmit_Click(object sender, EventArgs e)
    {
        Akismet api = new Akismet(ConfigurationManager.AppSettings["wordpressKey"], "", "Test/1.0");
        if (!api.VerifyKey()) throw new Exception("Key could not be verified");

        AkismetComment comment = new AkismetComment();
        comment.Blog = "";
        comment.UserIp = Request.UserAgent.ToString();
        comment.UserAgent = Request.UserAgent.ToString();
        comment.CommentContent = Server.HtmlEncode(txtComment.Text);
        comment.CommentType = "comment";
        comment.CommentAuthor = Server.HtmlEncode(txtName.Text);
        comment.CommentAuthorEmail = Server.HtmlEncode(txtEmail.Text);
        comment.CommentAuthorUrl = Server.HtmlEncode(txtWebsite.Text);

        DocumentType commentDt = DocumentType.GetByAlias("BlogPostComment");
        Document newComment = Document.MakeNew(txtName.Text, commentDt, new umbraco.BusinessLogic.User(0), Node.GetCurrent().Id);

        newComment.getProperty("name").Value = Server.HtmlEncode(txtName.Text);
        newComment.getProperty("email").Value = Server.HtmlEncode(txtEmail.Text);
        newComment.getProperty("website").Value = Server.HtmlEncode(txtWebsite.Text);
        newComment.getProperty("comment").Value = Server.HtmlEncode(txtComment.Text);

        newComment.Save();

        if (!api.CommentCheck(comment))
        {
            newComment.Publish(new umbraco.BusinessLogic.User(0));
            umbraco.library.UpdateDocumentCache(newComment.Id);
            Response.Redirect(umbraco.library.NiceUrl(Node.GetCurrent().Id) + "#comment-" + newComment.Id.ToString());
        }
        else
        {
            customError.IsValid = false;
            customError.ErrorMessage = "Your comment has been detected as spam.";
        }           
    }

I have also thought about creating a new button datatype to enable me to submit unblocked spam and Ham (false positives) back to Akismet to help improve the algorithm, but have not gotten around to it and really haven't needed to. Akismet really does a great job for lazy folks like me. Their claim of 93% accuracy is looking pretty solid.

Now all I need is for Umbraco to implement a "Delete at" feature for documents so that I can set an expiration date for all of the unpublished spam nodes that I have in my content tree.

The first project is NCEExam.com, an online study and testing system for the National Counselor Exam. This is version 3 of the system and is the first version to run on Umbraco (V4). There were a lot of unique challenges in this project such as securing FLV videos and downloadable content, as well as a custom shopping cart and e-commerce interface.

We will probably convert our shopping cart to Commerce4Umbraco eventually. However, it had not even reached the alpha stage when this site launched.

http://www.nceexam.com

The second project is Tutt & Daggs, a company specializing in workplace training and motivation for educational institutions, large corporations and non-profit organizations.

A couple of neat features were incorporated into this project. I utilized a JQuery slider on the bios page along with a Light Box gallery for promotional photos. Thanks to Umbraco, the client has no trouble maintaining this site on their own.

http://www.tuttdaggs.com

It is always interesting to see how others are using Umbraco in creative or innovative ways. If you have any projects you would like to share, please pass them along.

For example, if you use SyntaxHighlighter, you could use up to 21 brush scripts, plus the core script and theme style sheet. That makes for a lot of references in the head tag.  That is why I almost always add the following inside the <head> tag in my top lever master page between the style sheet and script references.

<asp:ContentPlaceHolder Id="head" runat="server">
<!-- extra references load from child -->
</asp:ContentPlaceHolder>

This allows me to specify any extra/temporary css or script references in the child template. In this example I even passed in the javascript function to generate a lightbox.

<asp:Content ContentPlaceHolderID="head" runat="server">
    <link href="/css/temp.css" rel="stylesheet" type="text/css" />
    <script type="text/javascript" src="/scripts/jquery.lightbox-0.5.min.js"></script>
    <script type="text/javascript">
    $(function() {
        $('#gallery a').lightBox();
    });
    </script>
</asp:Content>

I know this isn't groundbreaking, but it is a really great tip for anyone just starting out with Umbraco or master pages.

The good news is that Umbraco already uses UrlRewritingNet.UrlRewrite to create friendly URLs, so everything we need to get this working is already in place. After brushing up a bit on regular expressions I came up with the following two step solution.

First, I added the following filter to /config/UrlRewriting.config.

<add name="tagfilter"
  virtualUrl="^/blog/tag/(.*).aspx"
  rewriteUrlParameter="ExcludeFromClientQueryString"
  destinationUrl="/blog.aspx?filterBy=$1"
  ignoreCase="true"
/>

Then I modified BlogTags.xslt to generate the correct URL for each tag.

<a href="{link}/blog/tag/{current()}.aspx"><xsl:value-of select="current()"/> (<xsl:value-of select="@nodesTagged"/>)</a>

There you have it. Now my tag URLs are all search engine friendly and it only took about two minutes!

If you are a .NET developer and you have never heard of Umbraco, you really should check it out. In fact, this entire site is built on Umbraco and a modified version of the Blog for Umbraco 4 package by Tim Geyssens.

If you are familiar with Umbraco and need to get a blog up and running, I highly recommend starting with Blog for Umbraco 4. It will save you a lot of time and energy. You can find it in the Umbraco Package Repository.

I made some pretty major modifications to the Document Types, Master Pages, CSS, JS and XSLT. I also created a new comment form that better suited my needs. However, this was all extremely easy thanks to the flexibility of Umbraco and Tim's rather open architecture. Tim, if you ever read this, thanks a million for sharing!

One modification I made that I would like to pass along is in the XSLT for listing blog comments. Even though the current XSLT in the package specifies a height and width for the gravatars, these parameters are being ignored by the gravatar API because the XSLT is generating double ampersands between them.

As you can see below, I wrote what I consider to be a pretty ugly workaround.

<xsl:text disable-output-escaping="yes">
  <![CDATA[<img class="gravatar" src="]]>
</xsl:text>
<xsl:value-of select="BlogLibrary:getGravatar(data [@alias = 'email'], 30, '{your default image}')" disable-output-escaping="yes"/>
<xsl:text disable-output-escaping="yes">
  <![CDATA[" alt="Gravatar of ]]></xsl:text>
<xsl:value-of select="data [@alias = 'name']" disable-output-escaping="yes"/>
<xsl:text disable-output-escaping="yes">
  <![CDATA[" />]]>
</xsl:text>

CDATA everywhere!

If anyone has a better solution, I am all ears!

UPDATE! Please see the solution to this problem from Benjamin Howarth in the comments.